Shirin Ghaffary | Vox Our world has too much noise and too little context. Vox helps you understand what matters. 2023-08-01T15:37:42+00:00 https://www.vox.com/authors/shirin-ghaffary/rss https://platform.vox.com/wp-content/uploads/sites/2/2024/08/vox_logo_rss_light_mode.png?w=150&h=100&crop=1 Shirin Ghaffary <![CDATA[Why Meta is giving away its extremely powerful AI model]]> https://www.vox.com/technology/2023/7/28/23809028/ai-artificial-intelligence-open-closed-meta-mark-zuckerberg-sam-altman-open-ai 2023-08-01T11:37:42-04:00 2023-07-28T06:00:00-04:00

Last week, Meta made a game-changing move in the world of AI. 

At a time when other leading AI companies like Google and OpenAI are closely guarding their secret sauce, Meta decided to give away, for free, the code that powers its innovative new AI large language model, Llama 2. That means other companies can now use Meta’s Llama 2 model, which some technologists say is comparable to ChatGPT in its capabilities, to build their own customized chatbots. 

Llama 2 could challenge the dominance of ChatGPT, which broke records for being one of the fastest-growing apps of all time. But more importantly, its open source nature adds new urgency to an important ethical debate over who should control AI — and whether it can be made safe.

As AI becomes more advanced and potentially more dangerous, is it better for society if the code is under wraps — limited to the staff of a small number of companies — or should it be shared with the public so that a wider group of people can have a hand in shaping the transformative technology?

Top tech companies are taking different approaches

In Meta’s Llama 2 announcement, Mark Zuckerberg posted an Instagram of himself smiling with Microsoft CEO Satya Nadella, announcing the two companies’ partnership on the release. Zuckerberg also made the case for why it’s better for leading AI models to be “open source,” which means making the technology’s underlying code largely available for anyone to use.

“Open source drives innovation because it enables many more developers to build with new technology,” wrote Zuckerberg wrote in a separate Facebook post. “It also improves safety and security because when software is open, more people can scrutinize it to identify and fix potential issues.”

The move is being welcomed by many AI developers, researchers, and academics who say this will give them unprecedented access to build new tools or study systems that would otherwise be prohibitively expensive to create. Cutting-edge large language models like the ones that power ChatGPT can cost tens of millions of dollars to create and maintain.

“I’m just bracing myself for what kind of progress can happen,” said Nazneen Rajani, research lead at open source AI platform Hugging Face, which collaborated with Meta on the release. Rajani wrote a post on Twitter assessing Llama 2’s capabilities when it first came out and told Vox, “We will be able to uncover more secret ingredients about what it actually takes to build a model like GPT-4.”

But open-sourcing AI comes with major risks. Some of the biggest players in the field, including Microsoft-backed OpenAI and Google, have been limiting how much of their AI systems are public because of what they cite as the grave dangers of these technologies. 

Some technologists are increasingly worried about hypothetical doomsday scenarios in which an AI could outsmart human beings to inflict harm like releasing a biological super weapon or causing other havoc in ways we can’t fully imagine. OpenAI’s co-founder, Ilya Sutskever, told The Verge in February that his company was “flat-out wrong” when it shared details about its models more openly in the past because if AI becomes as intelligent as humans one day, reaching what some call AGI or artificial general intelligence, it would be unwise to share that with the masses. 

“If you believe, as we do, that at some point, AI — AGI — is going to be extremely, unbelievably potent, then it just does not make sense to open-source. It is a bad idea,” Sutskever said at the time.

While we may be far off from AIs that are capable of causing real human destruction, we have already seen AI tools from the open source community be misused in other ways. For example, soon after Meta released its first Llama model strictly for research use in February, it leaked on the anything-goes online message board 4Chan, where it was then used to create chatbots that spewed hateful content like racial slurs and, in some cases, scenes of graphic violence. 

“We take these concerns seriously and have put a number of things in place to support a responsible approach to building with Llama 2,” wrote Ahmad Al-Dahle, VP of generative AI at Meta, in an email to Vox. Those measures include “red-teaming,” or pressure-testing the model before its release by feeding it prompts expected to generate a “risky output,” such as ones about criminal conduct and hateful content, Al-Dahle said. Meta also fine-tuned its model to mitigate against this kind of behavior and put out new guidelines barring certain illegal and harmful uses.

Meta says it will continue to fine-tune its model for safety after its release. 

“When technology is released and refined in the open, we believe it ultimately leads to more transparent discussions, increased responsiveness to addressing threats, and increased iteration in building more responsible AI tools and technologies,” Al-Dahle said. 

Some experts point out, for example, that we had the problem of misinformation even before AI existed in its current form. What matters more at this point, they say, is how that misinformation is distributed. Princeton computer science professor Arvind Narayanan told Vox that “the bottleneck for bad actors isn’t generating misinformation — it’s distributing it and persuading people.” He added, “AI, whether open source or not, hasn’t made those steps any easier.”

To try to contain the spread of misinformation, companies creating AI models can put some restrictions on how their programs can be used. Meta, for example, has some rules barring users from using Llama 2 to promote violence or harassment, but those rules will likely prove difficult to enforce. 

It’s also worth noting that Llama 2 also isn’t fully open. Meta didn’t release the training data used to teach the latest model, which is a key component of any AI system; researchers say it’s crucial to measuring bias in AI systems. Lastly, Meta requires companies with over 700 million monthly users — so basically, only a handful of fellow tech giants like Google — to ask Meta’s permission before using the software.

Still, overall, Llama 2 is the most open sourced AI project we’ve seen recently from a major tech company. Which brings up the question of how other companies will respond.

So what exactly is the case for and against a more open sourced AI world? And what direction do we seem to be moving toward, especially given Meta’s recent announcement?

Open source can lead to more innovation

If you’re a casual user of AI tools like ChatGPT, you may not see the immediate benefits of open-sourcing AI models. But if you’re an AI developer or researcher, the introduction of open source LLMs like Llama 2 opens up a world of possibilities.

“It’s a huge deal,” said Anton Troynikov, a co-founder and head of technology of AI startup Chroma which builds databases that developers plug into AI systems to customize it with their data, facts, and tools. 

For someone like Troynikov, using Llama 2 could allow the company to give its users more control over how its data is used.

“Now you don’t have to send any data outside of your system, you can run it 100 percent internally on your own machines,” said Troynikov, who gave the example of doctors who don’t need to expose patients’ medical records out to a third party. “Your data no longer has to go anywhere to get these fantastic capabilities.”

Troynikov said he’s personally just started using Llama 2 and is still testing how well it works with his company’s technology.

It’s too early to see exactly how else Llama 2 will be used, but Meta’s Al-Dahle said it sees a “range of possibilities in the creation of chat-based agents and assistants that help improve productivity, customer service, and efficiency for businesses that may not have been able to access and deploy this technology otherwise.”

There’s also a self-interest here for improving Meta’s own products. If Meta puts its AI models into the wild, the open source community of outside engineers will improve its models, which Meta can then use to build the in-app AI tools that the company has said it’s working on, like business assistant chatbots. 

This way, Meta doesn’t have to put all of its resources into catching up to OpenAI and Google, which are further along in putting generative AI tools in their main product line.

Open-sourcing AI will tap into the “intelligence of the masses” 

Some leading experts think that if AI models are open sourced, they could become smarter and less ethically flawed overall. 

By open-sourcing AI models, more people can build on them and improve them. The open source AI company Stability AI has already created a model called “FreeWilly” that builds on top of Llama 2. It quickly became popular and can now outperform its genesis, Llama 2, in some tests. That has led it to rise to the top of Hugging Face’s leaderboard open source AI models. 

“People outside Meta are beating Meta at its own performance and its own models that they carefully collected and curated over the years. They were able to do it in a week,” said Rajani. “It’s very hard to beat the intelligence of the masses”

Meanwhile, the AI community has a strong history of open-sourcing knowledge. Google built and publicly shared the transformer model, which is a neural network that understands context, like language, by tracking relationships in between parts of data, like the words in a sentence. The model has become foundational in cutting-edge AI models, and is used in many applications including in ChatGPT (the “T” in GPT stands for transformer).

Open source models allow researchers to better study the capabilities and risks of AI and to stop the concentration of power in the hands of a few companies, Princeton professor Arvind Narayanan said, pointing out the risk of a technological “monoculture” forming.  

“Monoculture can have catastrophic consequences,” he said. “When the same model, say GPT-4, is used in thousands or millions of apps, any security vulnerability in that model, such as a jailbreak, can affect all those apps.”

Historically, experts point out, AI has blossomed as a field because company researchers, academics, and other experts have been willing to share notes.

“One of the reasons why data science and AI is a massive industry is actually because it’s built on a culture of knowledge sharing” said Rumman Chowdhury, co-founder of Humane Intelligence, a nonprofit developing accountable AI systems. “I think it’s really hard for people who aren’t in the data science community to realize how much we just give to each other.”

Moreover, some AI academics say that open source models allow researchers to better find not just security flaws, but more qualitative flaws in large language models, which have been proven to perpetuate bias, hallucinations, or other problematic content. 

While companies can test for some of these biases beforehand, it’s difficult to anticipate every negative outcome until these models are out in the wild, some researchers argue.

“I think there needs to be a lot more research done about to what point vulnerabilities can be exploited. There needs to be auditing and risk analysis and having a risk paper … all of these can only be done if you have a model that is open and can be studied,” said Rajani.

But open source AI could also go terribly wrong

Even the most ardent supporters of open AI models acknowledge there are major risks. And exactly how AI could go wrong runs the spectrum from more easily faking people’s identities to wiping out humanity, at least in theory. The most pressing argument in this scenario is that if AI does reach some kind of artificial general intelligence, it could then one day outsmart humans in ways we won’t be able to control.

In a recent senate hearing, OpenAI CEO Sam Altman told Congress that with “all of the dangers of AI, the fewer of us that you really have to keep a careful eye on — on the absolute, bleeding edge capabilities,” the easier it is for regulators to control. 

On the other hand, even Altman acknowledged the importance of allowing the open source community to grow. He suggested setting some kind of limit so that when a model meets certain “capability thresholds” for performing specific tasks, it should be forced to get a license from the government.

That’s one on which some proponents of open source seem to agree with Altman. If we reach the point when AI models get close to overtaking humanity, then maybe we can pump the brakes on open source.  

But the challenging question with AI is at what point do we decide that it’s too powerful to leave unfettered? And if the genie is out of the bottle at that point, will it be impossible to stop the progress of AI? Those questions are impossible to answer with certainty right now. But in the meantime, open source AI is here, and while there are real immediate risks, as well as ones that could snowball down the road, there are also clear benefits for all of us in having a wider group of people thinking about it.

]]> Shirin Ghaffary <![CDATA[Elon Musk is just throwing money at popular Twitter users now]]> https://www.vox.com/technology/2023/7/14/23794456/elon-musk-ad-revenue-share-creator-twitter-users 2023-07-14T12:18:33-04:00 2023-07-14T08:55:00-04:00
Elon Musk at the VivaTech conference in Paris in June 2023. | Nathan Laine/Bloomberg via Getty Images

“Breaking: Twitter revenue share has begun!” tweeted anti-Trump commentator Ed Krassenstein on Thursday afternoon, who followed up with a screenshot of $24,877 that he said had been deposited in his account. His twin brother Brian Krassenstein, who is also a political commentator and Twitter persona, similarly tweeted that he’d received $24,305 from Twitter. 

Back in February, Elon Musk promised he would start paying creators a share of Twitter’s ad revenue, and more recently, the billionaire said the initial chunk of payouts would amount to $5 million total. The Krassensteins’ posts were among several enthusiastic tweets from a select group of Twitter users who said on Thursday they’ve started getting tens of thousands of dollars in payouts from the app, seemingly based on the number of impressions — or views — they’re getting on replies to their tweets. The first batch of payments aren’t just for one month, but a backlog of the past five and a half months since Musk first promised he’d share a percentage of ad revenue with creators. 

Here’s how it works: Twitter shares an undisclosed percentage of the ad revenue it gets from replies to people’s tweets, directly to the user. So the more people reply to a user’s tweets and ads in those replies get viewed, the more money you’ll get. For now, the only users getting paid are ones who meet specific criteria. They must be a verified user — meaning, they pay for a blue check mark or have been gifted one — have 5 million impressions, or views, on posts in each of the last three months, and have a Stripe account linked to their Twitter account.

All this means most users won’t be getting paid, and even those who do may not be getting paid as much in the future. Some form of ad revenue-sharing has become relatively standard for social media platforms. YouTube, for example, pays its creator partners 55 percent of ad revenue for regular videos and 45 percent for YouTube shorts.

But $5 million in a batch of ad payouts, and millions more if it continues, is a significant amount of money to spend, especially considering Twitter has allegedly been failing to pay some of its office expenses and bills. As of last month, the company was facing eviction in its Boulder, Colorado, office for not paying rent.

So why is Twitter spending money on this new creator revenue program?

Because it needs the loyalty of creators, who attract eyeballs and advertising dollars to the platform. And these days, especially with the viral launch of Instagram Threads last week, creators have a lot of options about where to go. 

“They’re competing with the likes of big companies with very deep pockets,” said social media consultant Matt Navarra. That means Twitter wants to keep creators on its platform, especially as Threads becomes more popular. The Meta-owned Twitter-killer app amassed a record breaking 100 million users within a week of launching, making it one of the fastest growing apps ever. In Navarra’s words, Twitter has “no choice but to create such a program.” 

But since it financially rewards creators who have more replies, Twitter’s new program could incentivize its creators to post controversial tweets that spark heated conversation. One user pointed this out, tweeting, “The more haters you have in your replies the more money you’ll make on Twitter.” To which Musk replied, “Poetic justice.”

At the same time that Twitter is trying to incentivize creators to use the platform, it’s also trying to convince major brands to keep spending money advertising on the platform. And incentivizing Twitter users to rile up angry, reply-guy tweets even more than they already do might make Twitter an even more volatile place, and one that advertisers may be less likely to want to associate with.

“I’m not sure how good that is in terms of making the platform a happier, safer, lovely place to be around,” Navarra told Vox.

Advertisers cut back on spending on Twitter by 59 percent year over year in the first week of April, according to an internal document seen by the New York Times. This was reportedly because of what they see as a risk of harming their brand reputation if their ads appear next to controversial content.

So far, many of the creators who have said they’ve benefited from Twitter’s new ad revenue share program are independent journalists and political pundits. Aside from the Krassenstein brothers, other names include right-leaning conservative commentator Benny Johnson and podcaster Tim Pool.

For a long time, journalists essentially gave their content to Twitter for free, in exchange for attention on their articles and access to real-time information. Now, with many journalists tweeting less, these payouts seem aimed at hanging on to the journalists who remain, or the new creators who are emerging.  

“Next generation of journalists should be able to make a living doing it on Twitter. After many years, great to see it becoming possible …” Twitter product VP Keith Coleman said in a tweet on Thursday. 

Maybe Twitter will successfully create a new class of journalists through this program. But they’ll likely be a very different set of journalists than the ones who came up on the pre-Elon Twitter. And for the new generation of journalist creators, there’s some cash to be made on Twitter, at least for now. Whether Twitter will have the cash to keep paying these creators tens of thousands of dollars in payouts every few months — or even more often — has yet to be seen. 

]]> Shirin Ghaffary <![CDATA[Meta is finally having its moment with the Twitter alternative Threads]]> https://www.vox.com/technology/2023/7/10/23788945/meta-mark-zuckerberg-elon-musk-threads-twitter-instagram-social-media-platform 2023-07-10T12:59:16-04:00 2023-07-10T09:15:00-04:00
In a week, Meta’s social media app Threads, positioned as a competitor to Twitter, has reached 100 million users. | Rafael Henrique/SOPA Images/LightRocket via Getty Images

The much-anticipated wrestling cage match between Elon Musk and Mark Zuckerberg hasn’t happened yet, and may never happen at all.

But the bigger fight — the one about power, cultural relevance, and ego, played out on social platforms rather than in a steel cage — is very much still on. And after the events of the past few days, it’s clear that Zuckerberg has won the first round.

In just a few days, Meta’s new Twitter-killer app, Threads, has proven itself as the most viable contender to actually replace Twitter that has yet emerged. The app has grown so much and so quickly that it’s surprised everyone, including Zuckerberg himself. It is now possibly one of the most downloaded apps ever. 

“70 million sign ups on Threads as of this morning. Way beyond our expectations,” wrote the tech CEO in a thread on Friday. By Monday, less than a week after launch, it had surpassed 100 million users.

We don’t know if Threads’ millions of users will stick around — that’s a very big “if”  — but for now, there’s no denying that its launch represents a massive success for Meta. And Zuckerberg is basking in the glory. After 11 years of not tweeting, the Meta CEO turned to his rival’s app to troll Musk by posting a Spiderman clone meme on Wednesday. Musk has, in turn, punched back. Musk’s lawyers threatened a lawsuit accusing Meta of stealing Twitter’s trade secrets by hiring employees who had access to confidential information, while his recently appointed Twitter CEO — Linda Yaccarino — tweeted that the “Twitter community can never be duplicated.” As for Musk himself, he’s tweeted through it, posting a steady stream of digs at Zuckerberg and his apps.

“It is infinitely preferable to be attacked by strangers on Twitter, than indulge in the false happiness of hide-the-pain Instagram.” Musk replied to a tweet on the eve of Threads’ launch, taking aim at what Meta has positioned as the “friendly” place it wants Threads to be, unlike its notoriously snarky blue bird rival.

Musk’s dig gets at the core of one of Meta’s biggest hurdles with Threads: its potential to turn into a snoozefest if it doesn’t encourage authentically spicy posts over the bland marketing from professional brands and celebrities that tends to populate Instagram, as my colleague Rebecca Jennings wrote. So far, my own feed has been a mix of some cringe and some genuinely entertaining celeb posts (Martha Stewart’s age-embracing selfies being a good example), as well as content from lesser-known users sprinkled in the mix. One thing Instagram has said it’s not going to be: a place that encourages hard news and politics, per Instagram head Adam Mosseri’s posts on Friday. That could make Instagram a “less angry” place, as Mosseri wrote, but could also risk making Threads more boring than the vibrant public square of world-changing ideas Twitter was at its peak.   

But for now, the numbers don’t lie: millions of people — growing by the hour — are excited about Threads.  Out of all the many Twitter competitors that have gained attention in the past few months, like Bluesky, Mastodon, and Post News, Threads is the one that really stands a chance. 

Threads is the comeback win Meta needs

While Zuckerberg has struggled to convince the public to buy into his metaverse ambitions for the past two years, he has found success again by sticking to the basics of Meta’s business: social media. With Threads, Meta didn’t have to reinvent the wheel, which, historically, has not always been its strong suit. Instead, it did what it has done so well in the past with Instagram Stories and other copycat features: unabashedly cloning its competition with precision and scale that only it can bring to bear. And this time, Zuckerberg struck when a time when his opponent was particularly weak, as Twitter is facing a user backlash for temporarily limiting how many tweets users could see. Although Musk has said in the past few months that Twitter’s user numbers were growing, third-party reports suggest otherwise, showing that its traffic was trending downward by over 7 percent year over year as recently as April. 

Meanwhile internally at Meta, Threads is providing a much-needed morale boost following several months of recent mass layoffs, according to several sources who work at the company — with internal feedback groups and message boards buzzing.

Depending what happens next, Threads could be the beginning of a sustained comeback for Zuckerberg, who spent last year pushing a “year of efficiency” in hopes of rebounding Meta from a period of historic stock price dips and stalled user growth. Now, with stock prices gaining in the past few months and Threads generating something that has long been missing in Menlo Park — buzz — Zuckerberg’s staff has something to celebrate.

One Meta employee, who spoke on the condition of anonymity for fear of professional repercussions, said they were happy for Meta’s success and the initial stock bump following Threads’ release. Personally, they were also excited about an alternative to what they saw as Musk’s chaotic reign at Twitter. Previously, Meta chief product officer Chris Cox reportedly said in an internal meeting that Threads would be a more “sanely run” Twitter.

It’s a sentiment shared by many social media users outside the company who were eager to join Threads because of their growing frustration toward Twitter, an app that they view as increasingly leaning to the right under the fiat of a volatile CEO.

But other longtime Meta critics have argued that users have no reason to put more trust in the company. Some pointed out that Threads collects a wide set of user information, such as location, health and fitness information, and browsing history — although Twitter seemingly tracks a similar set of information based on details in the Apple app store.

That’s not surprising, because whatever platform it operates on, Meta runs its business by collecting data about us, its users, to better target the posts and ads it thinks we’ll click on (although Threads doesn’t have ads for now). And there have been times in the past when Meta has damaged that trust as guardians of our personal data, from how it allowed users’ data to be harvested for political purposes with the Cambridge Analytica scandal, to breaking EU regulatory guidelines on protecting teenagers’ user privacy on Instagram.

Still, many Meta negative-to-ambivalent users set aside whatever qualms they have to join Threads because it was a clear choice against Twitter. 

The result is that Zuckerberg — who just less than a year ago was the brunt of meme jokes about his disappointingly basic and legless metaverse avatar — has suddenly become the cool guy in corners of the internet.

One viral meme showed a photoshopped Zuckerberg posing gleefully next to Musk’s grave. Others showed Zuck swinging at Musk in a fight, or a distraught-looking Ben Affleck, labeled as Musk, anxiously smoking a cigarette. 

For Zuckerberg, it was a calculated risk to go after the territory of Musk, who has a troll army of diehard supporters by his side, ready to taunt Zuckerberg if he fails with Threads. But the risk was one worth taking. With Threads’ massive success, Zuckerberg is coming out on top, at least for now. After perhaps the darkest year in his nearly two decades-long reign, he has a real win.

]]> Shirin Ghaffary <![CDATA[What you need to know about Threads, Instagram’s new Twitter killer app]]> https://www.vox.com/technology/2023/7/5/23785140/threads-instagram-meta-twitter-killer-mark-zuckerberg-elon-musk 2023-07-06T11:46:27-04:00 2023-07-05T19:05:43-04:00
Meta CEO Mark Zuckerberg watching a UFC Fight Night event in October 2022. | Jeff Bottari/Zuffa LLC

Instagram’s much-hyped new Twitter-killer app, Threads, is here. And it looks a lot like Twitter. Which is exactly the point. 

Many social media users are ready — desperate even — for a solid Twitter replacement, as the app has been going through a particularly rough phase in what many see as its gradual product degeneration under Elon Musk’s leadership. Last weekend, the company started limiting the number of tweets people can read, a questionable business decision that was widely unpopular with users. While there are some alternatives out there, like Mastodon and Bluesky, none have grown to surpass Twitter’s popularity with a critical mass of politically and culturally influential figures.

So Meta-owned Instagram decided to strike while the iron is hot. The Threads app was initially expected to launch later this month, only to be moved up to this Thursday, and now, to today. The app will start to go live for users in 100 countries, although reportedly not in the European Union (more on that later).

“Our vision with Threads is to take what Instagram does best and expand that to text, creating a positive and creative space to express your ideas,” Instagram’s parent company, Meta, wrote in a company blog post on Wednesday.

Functionally, Threads is similar to Twitter, with some minor differences. You can write short posts of up to 500 characters that include links, photos, and short videos up to five minutes in length. Your Threads feed will be algorithmic, which means it will be populated by a mix of people you follow and recommended content; much like Instagram now. Twitter gives you the option to toggle between an algorithmic and chronological-based feed of only people you follow. But overall, based on early screenshots of the app shared with Vox, the apps look and feel quite similar.

The main defining feature that separates Threads from Twitter is that it has decentralized ambitions. Meaning that in the future, you should be able to plug your Threads posts into other social media platforms like Mastodon — a very different approach than Twitter, which has been limiting free API access to third-party developers. But the interoperability isn’t ready yet, according to Meta. And it’s also not what many day-to-day users care most about, which is: who’s posting on it and how easy it is to use. 

So how will this new app actually work, and what’s it like? And does it stand a real chance to overtake Twitter? 

How to use it and what it’s like

To use Threads, you’ll have to download it as a standalone app in the Apple or Android store. 

Once you have the app, you can log in with your Instagram account, and choose to follow the same people you already follow on that platform. This is one of Threads’ biggest advantages over other Twitter replacement apps: Over 2 billion people already have a built-in social network on Instagram, so unlike with, say, Mastodon, you don’t have to completely recreate your follower base from scratch.

The Instagram and Threads worlds are very much interconnected. If you’re verified on Instagram (which you can now pay for), that verification will roll over into Threads. And you can cross post your threads on Instagram as a story, or as a link to another platform. 

Once you’re in there, it functions a lot like Twitter, albeit with an Instagram design flair, including the same Instagram font and icons. You can like, reply, or repost a thread. The feed will be a combination of people you follow and recommended content from people you don’t follow, according to Meta. 

Getting the Threads feed algorithm right will be key for Instagram. Many users have complained about Twitter’s “For You” feed showing them too much content from random users they don’t want to see, and that they miss the old-school default-chronological feed on Twitter. We’ll see how users take to posts Threads thinks they want to see, versus those they’ve opted into.

What Threads’ decentralized approach means

Threads is the first app from Meta to push toward “decentralization” — the idea that users should be able to port their social media content, and interact with users, across different apps all built on the same underlying standards.

Mastodon is the most popular social network to run on a decentralized model, which advocates say can produce a better internet no longer dominated by a single social media company. Threads, similarly, plans to take a decentralized approach.

But it isn’t there yet.

Sometime “soon,” the company wrote in a blog post, Threads will be compatible with the ActivityPub protocol. It’s a system developed by the World Wide Web Consortium (W3C), an international organization that sets standards for the modern internet, to govern how social networks can run independently. 

The idea is that one day in the future, you can have your Threads posts be visible on other apps like Mastodon or WordPress, or vice versa, and have users comment on posts across the apps. And if you decided to stop using Threads altogether, you would hypothetically be able to port all your content over into a new app.

“We believe this decentralized approach, similar to the protocols governing email and the web itself, will play an important role in the future of online platforms,” wrote Meta in a blog post.

Decentralization is a buzzy concept in the tech world right now, and may give Threads appeal to the more digitally savvy audience. But most users aren’t familiar with decentralization, and probably don’t care about it too much. What will really matter is how many people end up downloading and actually liking their experience on the app, which gets us to our next point.

Regulatory concerns and other hurdles

Meta is going up against some significant regulatory and reputational hurdles when it comes to globally launching this app.

For example, Meta isn’t launching Threads in the EU for now because of regulatory uncertainty in the EU coming with the new Digital Markets Act, according to Bloomberg. The act limits what major companies designated as “gatekeepers” can do.

“Europe continues to be an incredibly important market for Meta. We are working on launching Threads in more countries and will continue to evaluate whether to launch in Europe, but the upcoming regulatory uncertainty has played into our decision not to launch right now,” a Meta spokesperson said in a statement to Vox.

Around privacy concerns, Meta said in its blog post that anyone under 16 (or 18 in certain countries) will default to a private profile when they join Threads. From a safety perspective, Instagram said it’s giving users the same tools that they have on Instagram to limit who can mention or reply to you, hide specific offensive words in replies, and unfollow, block, or restrict accounts.

But as the EU challenges show, Instagram will need to overcome something that a handful of privacy and safety features alone can’t change: underlying trust in its parent company, which has faced controversy about how it handles user data since the Cambridge Analytica scandal of 2018. 

Moreover, Threads will need to convince a critical mass of users that it’s not just trustworthy, but relevant. The magic of Twitter was that it’s a place where world leaders with immense power, snarky writers, A-list celebrities, and everyday very online users could all be in conversation with each other about the news of the day. For Threads to get that same effect, it will need those culture starters who can make compelling short, 500-character posts. 

Unlike Instagram proper, Twitter’s social currency is words, not pictures. Meta has been inviting major celebrities to join an early version of the app. Already, big names like Malala Yousafzai, Shakira, and Gordon Ramsay have used it, Meta confirmed. Threads perhaps stands the best chance of any Twitter competitor yet, and it will need more of those kinds of heavyweight figures whose words matter, and the users who follow them. 

]]> Shirin Ghaffary <![CDATA[The ad industry is going all-in on AI]]> https://www.vox.com/technology/2023/6/29/23777560/cannes-lions-google-meta-ai-advertising-2023 2023-07-10T16:05:22-04:00 2023-06-29T06:00:00-04:00
At Cannes Lions advertising festival in 2023, AI dominated the conversation. | Paige Vickers/Vox

“If you were branding this Cannes, it would be the AI Cannes,” Meta ad executive, Nicola Mendelsohn, told me last week. We were sitting in a glass-walled cabana on the French Riviera, steps away from the shimmering blue Mediterranean Sea. 

The Cannes she was referring to isn’t the one you’ve probably heard of — the film festival — but rather Cannes Lions, a similarly swanky festival celebrating advertising instead of cinema. 

Every June, thousands of advertising professionals fly in for a bonanza of events. While the festival’s official programming happens at the Palais des Festivals et des Congrès convention center, the real networking happens at beachside business meetings, yacht deck happy hours, and celebrity-studded after-parties. The hot-ticket items this year were Spotify’s invite-only concerts by Florence and the Machine and the Foo Fighters, consulting agency MediaLink’s and iHeartMedia’s exclusive Lizzo performance, and TikTok’s end-of-week closing party. On the iHeartMedia yacht, Paris Hilton DJ’ed to a crowd so packed that the party was shut down by the cops.

But it’s not all rosé and champagne: Cannes Lions is a high-stakes hustling opportunity for power brokers at tech companies, ad agencies, and consumer brands — think Nike, Unilever, and Coca-Cola — to check in on multimillion-dollar advertising deals in the second half of the year, and plan new ones for the year ahead.

This year, the festival came on the tail end of a particularly rough time for the tech and advertising world. Digital ad spending slowed down significantly in 2022 compared to years prior, primarily due to rising inflation, an unsteady global economy, and policy changes that made it harder to track users’ browsing habits. That decline contributed to mass layoffs and budget cuts across the media industry. Although conditions are improving a bit, it’s unlikely spending will return to the levels it reached in the early pandemic, and the latest forecasts show continued advertising spending cuts. Given the economic uncertainty, some companies were sending fewer staffers to the festival and cutting back on their presence.

But everyone wants a reason to party and make deals at Cannes Lions. Since advertising funds so many of the free online services we rely on — everything from Facebook to Google to media publishers, including Vox — the industry’s success or failure has massive effects on the average consumer. And in the past year, the advertising industry has desperately needed something to be optimistic about. 

Luckily for those looking for a vibe shift, AI had officially entered the chat.

For a week in June, the developing technology was the talk of the beach in the south of France. And while I’m used to nonstop AI hype back home in Silicon Valley, I was not expecting to experience so much of it in Cannes. The streets were plastered with billboards; panels and late-night party chatter were all about AI. Google demoed new tools, Meta announced an upcoming AI assistant that will help advertisers make ads, and Microsoft hosted back-to-back days of AI-themed programming at a beachside venue decorated with images of AI-generated sea creatures.

There was so much AI talk at Cannes Lions this year that, at times, people sounded sick of talking about it. “I’m trying to find the AI superpowered yacht,” I heard one attendee say in jest as he sat on the deck of a luxury vessel, drink in hand. 

Jokes and some healthy cynicism aside, the questions everyone seemed to be asking hint at some pretty serious shifts for the media business. Will AI fundamentally change the way we create and consume advertising? Will it be able to lift digital advertising out of its slump? And will it ultimately enhance or replace the human creativity that goes into making ads? Will it save (or destroy) journalism?

AI isn’t new, but it’s the savior the ad industry needs right now

Six years ago, one of the world’s largest advertising agencies, Publicis Groupe, was widely ridiculed for cutting its marketing presence at Cannes so that it could instead invest money into developing a new AI business assistant, called Marcel. Clients and competing ad firms alike dismissed the idea that AI was a worthwhile endeavor for an agency in the business of human creativity.

“At the time, it was panned by everybody, but now it looks pretty smart,” Jem Ripley, the US CEO of digital experience for Publicis, told me in the hotel lobby of the Le Majestic hotel, a hot spot for executive meetings at the conference. To rub it in a little, this year, Publicis launched a hate-to-say “I told you so” billboard campaign around Cannes reminding people how prescient they’d been with developing the AI-powered Marcel platform.

Even before they became hot buzzwords in the industry, automation and AI were powering advertising behind the scenes for years. The two biggest digital advertising platforms, Google and Meta, have long used AI technologies to develop the automated software that determines the price they charge for an ad, who they show the ad to, and even what lines of marketing copy are most effective to use. As users, we don’t see it day-to-day, but that technology is core to many tech companies’ businesses.

On the consumer side of things, apps like TikTok, Instagram, and YouTube all build AI into the underlying algorithms that decide what content you see, based on what the tech thinks you’re interested in. Think about how TikTok predicts what funny videos you want to see next or how Google ranks your search results; all of it uses AI. 

“Everybody wants this to be the year of AI, which I think to some degree it is,” said Blake Chandlee, TikTok’s president of global business solutions, sitting with me in his company’s Cannes outpost inside the swanky Carlton Hotel. “AI is not new. This concept of large language models, it’s been around for years. … What’s new is ChatGPT and some of the bots and the applications of the technology.”

Just as everyone from artists to writers has learned the value of AI from apps like ChatGPT, Midjourney, and Bard, advertising companies are now realizing what these tools can do for them. That mainstream adoption, combined with the fact that marketers are looking to cut costs in this uncertain economic climate, means that AI is exploding in the ad industry at this moment.

I chatted with everyone from creative directors at the top of the totem pole to rank-and-file copywriters at the festival last week, and almost everyone I spoke with said they had experimented with AI tools in their day-to-day duties. And not because their boss told them to, but because they thought it could save them time writing an email, sketching an ad mockup, or brainstorming an ad concept. Some of them were also worried that it could one day replace their jobs — more on that later — but for now, they were having fun with it.

“I think this year is particularly exciting because it’s sort of like the iceberg breaking through the surface,” said Vidhya Srinivasan, vice president and general manager for Google Ads, in an interview at Google’s beach outpost last Wednesday. “And so I think it’s more personal, and it’s much more tangible for people now. And that brings about a different kind of energy.”

What the AI future of advertising will look like

Standing onstage in a grand theater at the Palais du Festival, Robert Wong, vice president of Google Creative Lab, touted the AI tools his company has starting to put in the hands of advertisers. 

In one demonstration, Wong showed how a client can upload a single image of a company logo — a colorful Google “G” icon, in his demo — into Google’s systems and immediately get back a bunch of high-quality 3D images in the same branded style, from a Google dog cartoon to a Google-branded glass of rosé, which was fitting for the venue.

While this quick demo may not seem dramatic compared to some of the splashy generative AI creations we’ve seen lately, like the Pope in a puffer jacket, it was met with “oohs” and “ahhs” from the audience of advertising professionals. That’s because for designers, work like that could take days or weeks. In just a few keystrokes, this new Google tool could give them limitless iterations of a design to experiment with.

“Day-to-day, what I see is designers literally doing sketches in a matter of seconds versus hours. And not one, but like 10,” said Wong in a press conference after the presentation. “And that’s just the beginning. I think we don’t even know what these tools might be in the future.”

Meta also made some AI announcements at the conference, including that it’s working on an AI-powered assistant that can help advertisers create ads. With its so-called AI Sandbox, the company in May released a slew of advertising tools that let advertisers use quick text prompts to come up with AI-generated advertising copy, create different visual backgrounds for their ads, or resize their images. For now, the program is only open to a small group of beta testers, but it’s expanding to more users later this year. 

In the long run, the cost savings for brands using generative AI for advertising could be “substantial,” according to Mendelsohn, Meta’s global head of business group.

“It gets better as we train the machines,” she said during our interview at Meta Beach. “And then you think about the reduction not just in cost, but in the impacts on climate. People are not having to travel to be able to do shooting in different ways, or even the reusing of back catalog of ads and things in the past.”

As the tech giants build out tools for their advertising customers, some are already experimenting with open source generative AI software with some impressive results.

For example, some major household brands are already starting to use AI to create high-production-value commercial videos. 

In October, Coca-Cola enlisted the AI image creation tool Stable Diffusion to help create a video that was shortlisted for an award at the festival. The ad, called “Coca-Cola Masterpiece,” used AI in addition to traditional methods, like CGI, to create complex animations under a tight deadline. The two-minute spot shows characters popping out of the art in a gallery to toss a classic Coca-Cola bottle in and out of famous paintings, like a Warhol and a van Gogh; the bottle takes on the visual style of the work of art when it enters each picture. It’s an incredibly complex animation process that took only eight weeks, according to visual effects company Electric Theatre Collective, which Coca-Cola commissioned. Without the help of AI, the company told Digiday, it could have taken five times longer. 

“We wanted to use technology to get the kind of perfection we needed, the kind of quality we needed, in a short time,” Pratik Thakar, Coca-Cola’s global head of generative AI, said on a panel hosted by Microsoft.

Generative AI holds promise for creating new kinds of audio advertising, too. Spotify, for instance, is exploring whether it can train AI on specific people’s voices so that it can one day generate original audio ads from scratch.

“Can we start to get to a place where — I use Morgan Freeman as a canonical example — if you go and license the IP for his voice, can we use machines to help scale that even further?” said Lee Brown, global head of advertising for Spotify, which has been growing its ad business in recent years. “So is there an opportunity here for us? I think there’s a lot of potential there.” 

Some of these more sophisticated generative AI tools are still just possibilities for the ad industry at the moment. In the meantime, both Google and Bing are doing something a bit simpler: putting ads inside the conversations people are having with their AI chatbot assistants (Search Generative Experience and BingAI, respectively). The companies say this helps advertisers show users ads that are more relevant to people than what they’d see in a regular search.

The idea is that when you’re researching something like how to plan a trip to Greece, a chatbot would have more context about what you’re looking for — somewhere near the beach that’s kid-friendly in June, for instance — based on a series of follow-up questions you’re having with the bot rather than just through a single search query.

“From a marketer’s point of view, it’s interesting because you have a deeper insight into the user’s intent, because they’re in the conversation where you have more context about what they’re doing,” said Google’s Srinivasan.

A presenter onstage in front of a screen that reads “Human intelligence x artificial intelligence.”

In other words, with generative AI search engines, people ask detailed follow-up questions and actually talk to the bots. Jennifer Creegan, general Manager of global marketing and operations for Microsoft advertising, said in a panel last Wednesday that people’s search queries are three times longer in BingAI because of this back and forth. This leads people to click on an advertiser link, she added, and buy something more quickly.

“The best thing about all of this is this is not something I’m showing you in PowerPoint at Cannes to talk about the future,” Creegan said. “This is real. This is in the wild today. People are using it.” 

The concerns about AI and ads 

Even though new advancements in AI and advertising are real and in the wild, human judgment still needs to play a role in how it all works. Advertisers aren’t ready to fully hand over the reins to the robots to make their ads.

First of all, AI doesn’t replace taste. That means humans still need to review all the draft AI marketing copy or artwork manually. That’s because big companies are still cautious about protecting their brands, and it’s up to the people at the ad firms they hire to make judgment calls.

 “At the end of the day, there’s still a healthy concern — I think rightfully so — from our clients about what is going out there,” said Publicis executive Ripley. 

Another reservation major brands have around AI is that it could use other people’s creative work that it scrapes from the web, which could open them up to copyright infringement lawsuits. Publicis recently joined C2PA, a standard that watermarks images created by generative AI and can attach proper copyright information to it so that artists get credit for their work. 

Advertisers are also worried about brand safety. Given how AI chatbots have a propensity to generate incorrect information, also known as “hallucinations,” or occasionally veer off into emotionally loaded conversations, advertisers need to make sure that the quality of AI-generated ads is up to par.

“For every hour you put into generative AI as a business driver, you need to put an hour into governance,” said Lou Paskalis, a longtime ad executive who’s now chief strategy officer of Ad Fontes Media. “You need to make sure you don’t create a monster.”

All this raises some red flags for the workers in the ad industry. After all, if generative AI can reduce the number of people it takes to, say, produce a video or sketch an animation, the technology could wipe out a swath of jobs, particularly those on the creative side.

Among many advertising executives at Cannes Lions this year, there was an acknowledgment that AI will fundamentally change the kind of work people do. Despite tech companies’ optimism that AI will enhance and not replace human creativity, many said the new technology will get rid of some jobs while creating other new ones. One common refrain from ad execs was that the more creative your work is, the harder it will be to replace.

In the words of Coca-Cola’s Thakar, “Five-out-of-10” level creative advertising work is “free now.” He said, “So we need to figure it out … if you are really doing nine-out-of-10 work, then definitely there is always a demand.” 

Other executives compared AI to the invention of photography, which didn’t entirely replace painters as some thought it would. like Google’s SVP of research, technology, and society, James Manyika.

“AI and art are not at odds,” Manyika said in a keynote introducing Google’s new advertising tools. “AI doesn’t replace human creativity. It enhances, enables, and liberates it.” 

Ultimately, it doesn’t seem as though any of the concerns about AI stealing or replacing people’s work are stopping advertisers from jumping on the AI bandwagon. This embrace of the new technology could be a boon to the struggling ad industry. And that, in turn, could benefit consumers who rely on free services propped up by advertising. 

But like every other industry AI is impacting, the rise of AI-powered ads will force us to decide what still needs a human touch and what we’re happy to leave to the bots to handle.

]]> Shirin Ghaffary <![CDATA[What will stop AI from flooding the internet with fake images?]]> https://www.vox.com/technology/23746060/ai-generative-fake-images-photoshop-google-microsoft-adobe 2023-07-10T16:04:27-04:00 2023-06-03T07:00:00-04:00

On May 22, a fake photo of an explosion at the Pentagon caused chaos online.

Within a matter of minutes of being posted, the realistic-looking image spread on Twitter and other social media networks after being retweeted by some popular accounts. Reporters asked government officials all the way up to the White House press office what was going on.

The photo was quickly determined to be a hoax, likely generated by AI. But in the short amount of time it circulated, the fake image had a real impact and even briefly moved financial markets.

This isn’t an entirely new problem. Online misinformation has existed since the dawn of the internet, and crudely photoshopped images fooled people long before generative AI became mainstream. But recently, tools like ChatGPT, DALL-E, Midjourney, and even new AI feature updates to Photoshop have supercharged the issue by making it easier and cheaper to create hyperrealistic fake images, video, and text, at scale. Experts say we can expect to see more fake images like the Pentagon one, especially when they can cause political disruption. 

One report by Europol, the European Union’s law enforcement agency, predicted that as much as 90 percent of content on the internet could be created or edited by AI by 2026. Already, spammy news sites seemingly generated entirely by AI are popping up. The anti-misinformation platform NewsGuard started tracking such sites and found nearly three times as many as they did a few weeks prior. 

“We already saw what happened in 2016 when we had the first election with a flooding of disinformation,” said Joshua Tucker, a professor and co-director of NYU’s Center for Social Media and Politics. “Now we’re going to see the other end of this equation.” 

So what, if anything, should the tech companies that are rapidly developing AI be doing to prevent their tools from being used to bombard the internet with hyperrealistic misinformation? 

“We all have a fundamental right to establish a common objective reality”

One novel approach — that some experts say could actually work — is to use metadata, watermarks, and other technical systems to distinguish fake from real. Companies like Google, Adobe, and Microsoft are all supporting some form of labeling of AI in their products. Google, for example, said at its recent I/O conference that, in the coming months, it will attach a written disclosure, similar to a copyright notice, underneath AI-generated results on Google Images. OpenAI’s popular image generation technology DALL-E already adds a colorful stripe watermark to the bottom of all images it creates. 

“We all have a fundamental right to establish a common objective reality,” said Andy Parsons, senior director of Adobe’s content authenticity initiative group. “And that starts with knowing what something is and, in cases where it makes sense, who made it or where it came from.”

In order to reduce confusion between fake and real images, the content authenticity initiative group developed a tool Adobe is now using called content credentials that tracks when images are edited by AI. The company describes it as a nutrition label: information for digital content that stays with the file wherever it’s published or stored. For example, Photoshop’s latest feature, Generative Fill, uses AI to quickly create new content in an existing image, and content credentials can keep track of those changes.

AI-labeling tools like Adobe’s are still in their early stages, and by no means should they be considered a silver bullet to the problem of misinformation. It’s technically possible to manipulate a watermark or metadata. Plus, not every AI generation system will want to disclose that it’s made that way. And as we’ve learned with the rise of online conspiracy theories in recent years, people will often ignore facts in favor of believing falsehoods that confirm their personal beliefs. But if implemented well — and especially if these labels are seen as more neutral than traditional social media fact-checking — AI disclosures could be one of our only hopes for navigating the increasingly blurry distinction between fake and real media online.

Here is how some of these early AI markup systems could work, what the limitations are, and what users can do to navigate our confusing post-truth internet reality in the meantime.

The devil is in the metadata

When you look at an image on social media or a search engine today, odds are you don’t know where the photo came from — let alone if it was created by AI. But underneath the hood, there’s often a form of metadata, or information associated with the digital image file, that tells you basic details, like when and where the photo was taken. Some tech companies are now starting to add specific metadata about AI to their products at the moment of creation, and they’re making that information more public in an effort to help users determine the authenticity of what they’re looking at.

Google recently said it will start marking up images made by its own new AI systems in the original image files. And when you see an image in Google Search that’s made by Google’s AI systems, it will say something like “AI-generated with Google” underneath the image. Going a step further, the company announced it’s partnering with publishers like Midjourney and stock photography site Shutterstock to let them self-tag their images as AI-generated in Google Search. This way, if you come across a Midjourney image in Google Search, it will say something like “Image self-labeled as AI-generated”

Google Search public liaison Danny Sullivan said that this kind of AI labeling is part of a broader effort to give people more context about images they’re seeing.

”If we can show you a helpful label, we’re going to want to do that,” said Sullivan, “but we’re also going to want to try to give you background information that we can determine independent of the label.”

That’s why Google is also adding an “About this image” feature next to image search results — whether they are AI labeled or not — that you can click and see when the image was first indexed by Google, where it may have first appeared, and where else it’s been seen online. The idea is, if you searched for, say, “Pentagon explosion” and saw a bunch of images in the results, you would be able to see a fact-checked news article debunking the piece.

“These tools are really designed to help people understand information literacy more and bake it into the search product itself,” said Sullivan. 

Other major industry players have also been working on the issue of how to label AI-generated content. In 2021, a group of major companies including Microsoft, Adobe, the BBC, and Intel created a coalition called the C2PA. The group is tasked with helping to create an interoperable open standard for companies to share the provenance, or history of ownership, of a piece of media. C2PA created its first open standard last January, and since then, Adobe and Microsoft have released features using that standard.

For example, if you’re a photographer at a news outlet, you can mark when a specific picture was taken, who took it, and have that be digitally signed by your publisher. Later, your editor could make changes to the photo, signing it again with a seal of authenticity that it’s been verified by the C2PA standard. This way, you know that the photo was taken by a person — not generated by AI— and know who has made edits to it and when. The system uses cryptography to preserve the privacy of sensitive information.

“Now you can read the entire lineage of the history of a piece of digital content,” said Mounir Ibrahim, EVP of public affairs and impact at Truepic, a visual authenticity app that is a member of C2PA. “The purpose of us is to help content consumers … decipher the difference between synthetic and authentic.”

However flawed these early AI flagging and identification systems are, they’re a first step

Knowing the history and provenance of an image could potentially help users verify the legitimacy of anything from a headshot on a dating app to a breaking news photo. But for this to work, companies need to adopt the standard. 

Right now, it’s up to companies to adopt the C2PA standard and label verified content as they wish. The organization is also discussing potentially standardizing the look of the C2PA content credential when it shows up on images, Ibrahim said. In the future, the C2PA credential could be similar to the little padlock icon next to the URL in your browser window that signifies your connection is secure. When you see the proposed C2PA icon, you would know that the image you’re seeing has had its origins verified. 

So far, two big C2PA members, Adobe and Microsoft, have announced tools that integrate C2PA standards into their products to mark up AI-generated content. Microsoft is labeling all AI-generated content in Bing Image Generator and Microsoft Designer, and Adobe is using C2PA standards in its new AI Firefly product’s content credentials.

“The biggest challenge is we need more platforms to adopt this,” said Ibrahim. 

While the C2PA-style metadata labels work behind the scenes, another approach is for AI systems to add visible watermarks, as OpenAI has done with the rainbow bar at the bottom of DALL-E images. The company says it’s also working on a version of watermarking for its text app, ChatGPT. The challenge with watermarks, though, is that they can be removed. A quick Google search turns up forms of people discussing how to circumvent the imprint. 

Another imperfect option is technology that can detect AI-generated content after the fact. In January, OpenAI released a tool that lets you cross-check a block of text to determine whether it’s likely written by AI. The problem, though, is that by OpenAI’s own assessment, the tool is not fully reliable. It correctly identified only 26 percent of AI-written texts in OpenAI’s evaluations, although it’s notably more accurate with longer than shorter text. 

“We don’t want any of our models to be used for misleading purposes anywhere,” said a spokesperson for OpenAI in a statement. “Our usage policies also require automated systems, including conversational AI and chatbots, to disclose to users that they are interacting with our models.”

At the end of the day, even if these early AI flagging and identification systems are flawed, they’re a first step.

What comes next 

It’s still early days for tech platforms trying to automate the identification of AI-generated content. Until they identify a dependable solution, however, fact-checkers are left manually filling in the gaps, debunking images like the Pope in a puffy jacket or fake audio of politicians. 

Sam Gregory, executive director of human rights and civic journalism network Witness, who works with fact-checkers largely outside of the US, said that while he thinks technical solutions to AI identification like watermarking are promising, many fact-checkers are worried about the onslaught of misinformation that could come their way with AI in the meantime. Already, many professional fact-checkers are dealing with far more content to check than humanly possible. 

“Is an individual going to be blamed because they couldn’t identify an AI-generated image? Or is a fact-checker going to be the one to take the strain because they’re overwhelmed by this volume?” said Gregory. The responsibility to address AI misinformation “needs to lie on the people who are designing these tools, building these models, and distributing them,” he added.

In many cases, Gregory says, it’s unclear exactly what social media platforms’ rules are about allowing AI-generated content.

There is an urgency to figure out these problems as AI-generated content floods the internet

TikTok has one of the more updated policies around “synthetic media,” or media that is created or manipulated by AI. The policy, which was revised in March 2023, allows synthetic media but requires that, if it shows realistic scenes, the image must be clearly disclosed with a caption, sticker, or otherwise. The company also doesn’t allow synthetic media that contains the likeness of any private figure or anyone under 18. TikTok says it worked with outside partners like the industry nonprofit Partnership on AI for feedback on adhering to a framework for responsible AI practices.

“While we are excited by the creative opportunities that AI opens up for creators, we are also firmly committed to developing guardrails, such as policies, for its safe and transparent use,” a TikTok spokesperson said in a statement. “Like most of our industry, we continue to work with experts, monitor the progression of this technology, and evolve our approach.”

But many other platforms have policies that might need some updating. Meta, which owns Facebook and Instagram, and YouTube both have general rules against manipulated media that misleads users, but those could be clarified regarding what uses are acceptable or not, according to Gregory. Meta’s fact-checking policies state that manipulated media containing misinformation is eligible for fact-checking by its third-party partners, as it did with the fake Pentagon AI explosion claims.

“AI is bigger than any single person, company, or country, and requires cooperation between all relevant stakeholders,” Meta said in a statement. “We are actively monitoring new trends and working to be purposeful and evidence-based in our approach to AI-generated content.” 

Technological solutions to help people fact-check content themselves, like AI detection systems and watermarks, couldn’t come sooner.

But NYU’s Tucker says we need to test these solutions to see whether they’re effective in changing people’s minds when they encounter misleading AI content, and what the disclosures need to look to be impactful. For example, if the disclosures that an image or video is AI-generated are too subtle, people could miss it entirely. And sometimes, labels don’t work as expected. For example, Tucker co-authored a study last year showing that high- or low-quality news credibility labels had limited effects on people’s news consumption habits and failed to change people’s perceptions.

Still, there’s hope that if AI disclosures are seen not as politicized fact-checks but as neutral context about the origins of an image, they could be more effective. To know whether these labels are resonating with people and changing their minds will require more research.

There is an urgency to figure out these problems as AI-generated content floods the internet. In the past, tech companies had time to debate the hypothetical risks of AI misinformation because mainstream generative AI products weren’t yet out in the wild. But those threats are now very real. 

These new tools that label AI-generated content, while far from perfect, could help mitigate some of that risk. Let’s hope tech companies move forward with the necessary speed to fix problems that come with AI as quickly as they’re being created.

]]> Shirin Ghaffary <![CDATA[Mark Zuckerberg says the hardest part of Meta’s “year of efficiency” is over]]> https://www.vox.com/technology/2023/5/18/23729176/meta-silicon-valley-massive-layoffs-mark-zuckerberg 2023-05-26T01:46:14-04:00 2023-05-25T18:36:14-04:00

Meta laid off 5,100 more employees on Wednesday in its third round of mass cuts in the past three months, according to executive remarks in a company meeting on Thursday. That brings the company’s total layoffs to 10,600 in the first half of 2023, as part of Mark Zuckerberg’s planned “year of efficiency” to cut costs, shake up company culture, and narrow focus in response to slower growth in the tech industry. The company has also closed 5,300 open roles, Meta’s head of people, Lori Goler, told employees.

The company’s executives announced details about the layoffs on Thursday morning in a Q&A with employees that Vox obtained a recording of. 

“We’re saying goodbye to a lot of really talented people who’ve been part of this company,” said Zuckerberg in his opening remarks, before telling remaining staff that “the bottom line is that you’re here today.”

“I know that this has been a particularly thrashy period,” he said. “My hope is to provide as much stability moving forward as possible.”

A spokesperson for Meta declined to comment.

Zuckerberg has his work cut out for him to return Meta’s company culture to a state of normalcy after months of layoffs that have tanked employee morale, leaving many uncertain about their futures and some reportedly unsure of what to focus on. The CEO is hoping the worst is behind his company. While he wouldn’t rule out future layoffs, especially smaller ones, he said this was the last planned mass wave for now. Zuckerberg also said that employees would receive an update about return-to-office plans in the coming weeks giving more “consistent expectations and guidelines” around when and how often employees need to be in the office in person.

 ”We want to get a more of a critical mass of people in-person together in the offices a few days a week,” Zuckerberg said.

Meta’s continued downsizing is one of the starkest examples of how many major tech companies are tightening their belts after nearly two decades of uninterrupted growth. Silicon Valley as a whole has been going through an economic downturn that has caused major tech firms like Meta to drastically cut back on employee staffing and benefits. While Wall Street has responded positively to Meta’s cuts, the layoffs have taken a toll on Meta’s workforce.

Do you work at Meta and have thoughts about what’s going on at the company? You can send us a tip at shirin.ghaffary@protonmail.com. We can grant requests for anonymity if needed. Signal number available upon request.

“Restructuring is obviously a very difficult thing,” Zuckerberg said in response to a question about how Meta can rebuild its culture. “So it’s not like you can bounce back immediately. And in some ways, my goal has been to change our culture.”

He added that Meta could more easily shift to a “scrappier” work culture with a smaller staff.

“We were this big company, and I think we were getting a bit more bureaucratic. Part of the point of some of this restructuring is to break up the mode of it,” said Zuckerberg. “So yeah, I mean, some teams are maybe a little smaller now than [they] would be comfortable [with], and that causes issues in some ways for sure. But in other ways, I think it just forces us to find ways to be scrappy, or get things done more efficiently, and that means that there are going to be fewer environments or projects where there are too many cooks in the kitchen.”

From 2019 to 2022, Meta nearly doubled its headcount. But that’s when the company’s profits and user engagement were soaring. That started changing last February, when Facebook, for the first time, reported a decline in total users, and the advertising industry as a whole — the company’s main line of business — started to slow down. In recent months, some employees have openly questioned Zuckerberg and company leadership about whether they should be held accountable for decisions that led to the mass cuts.

Zuckerberg first announced in March that the company planned to eliminate 10,000 positions by the end of May, after previously cutting 11,000 in November. Last month, Meta cut around 4,000 of those planned 10,000 positions, leaving about 6,000 positions potentially on the chopping block this round. At the end of 2022, Meta, which is the parent company of Facebook, Instagram, and WhatsApp, had around 86,000 employees.

Last year was arguably the most brutal in Meta’s nearly 20-year history. Meta has been making somewhat of a comeback financially, however. It had stronger-than-expected earnings last quarter, and its stocks have recovered from historic lows. But now, Zuckerberg needs to make a comeback with his employee culture, too.

Update, May 18, 6:30 pm ET: This story, originally published on May 18, has been updated to include additional details about the layoffs after they happened.

]]> Shirin Ghaffary <![CDATA[Elon Musk tried a big presidential broadcast event and Twitter broke]]> https://www.vox.com/technology/2023/5/24/23735940/elon-musk-ron-desantis-twitter-tucker-carlson-2024-election 2023-05-24T19:56:29-04:00 2023-05-24T19:53:17-04:00
Elon Musk outside SpaceX’s launch facility in South Texas in February, 2023. | Jonathan Newton/Washington Post via Getty Images

Elon Musk’s high-profile Twitter event to launch the presidential bid of Florida Gov. Ron DeSantis on Wednesday got off to a rough start, crashing several times about 30 minutes into the broadcast before restarting the whole thing.

“That was insane, sorry,” said Musk after launching a new audio broadcast on Twitter Spaces about half an hour after the scheduled start time. “We’re actually doing this from David Sacks’s Twitter account because doing it from mine basically broke the system.”

Musk was referring to tech investor and entrepreneur David Sacks, who moderated the interview. Sacks claimed that this was the largest group that has “ever met online.” In reality, the number of people was far smaller than previous online events on or off Twitter. For example, more than 12 million people attended a virtual Travis Scott concert in Fortnite, and according to Twitter, more than 3 million people listened to Musk’s recent Twitter Spaces with the BBC.

During the interview, DeSantis said he chose to announce his presidential bid on Twitter in part because he aligns with Musk’s self-proclaimed free speech values. He also repeatedly criticized the so-called “woke mob” and “woke mind virus,” which is also a grievance of Musk’s.

“I think free speech in this in this country was on its way out the door,” said DeSantis. “That did not happen during Covid. Truth was censored repeatedly, and now that Twitter is in the hands of a free-speech advocate, that would not be able to happen again on this Twitter platform. So I think what was done to Twitter is really significant in the future of our country.”

While DeSantis claims he’s all about free speech, critics say that the Florida governor, like Musk, is actually serving a right-wing agenda that isn’t so free for everyone.

Musk has long claimed he wants Twitter to be a digital town square open to debate from all aspects of the political spectrum. 

“For Twitter to deserve public trust it must be politically neutral, which effectively means upsetting the far right and the far left equally,” the billionaire tweeted in April last year, shortly after he made his bid to buy the company. 

But lately, Musk has been upsetting one side a lot more than the other. He has been courting some of the most powerful figures in conservative politics to make Twitter their platform of choice, while angering liberals by engaging with conspiracy theories and culture-war-baiting rhetoric. 

That approach was clear on Tuesday, when Musk hosted DeSantis on Twitter. It was the first time a presidential candidate has announced their presidential bid on a social media network. It’s also notable that Musk, the company’s owner, is throwing his star power and massive following behind the effort. The DeSantis event happened the same day the Daily Wire, a conservative media outlet that hosts shows by popular right-wing pundits like Ben Shapiro and Matt Walsh, said it would be streaming its shows for free on Twitter. And just two weeks prior, recently fired Fox News host Tucker Carlson said he’s producing a new show that will run on Twitter — another major right-wing media coup for the platform.

While Musk has been busy promoting right-wing powerhouses on Twitter, he hasn’t made any similar public partnerships with liberal politicians, left-leaning or even neutral media outlets. His cozying up to the right seems to be alienating some liberal users. A recent Pew study shows that Twitter users who identify as Democrats were almost 10 percent more likely to say they would stop using the platform in a year (the partisan gap was even greater with Democratic women than men). And in the weeks after Musk took over Twitter, high-profile Republican Twitter accounts gained tens of thousands of followers while their Democratic counterparts experienced a decline, according to a Washington Post analysis.

It’s Musk’s prerogative to encourage whoever he wants on Twitter. While he says he will soon be stepping down as CEO and handing the position to former NBCUniversal ad executive Linda Yaccarino — who has her own conservative credentials as a well-known Trump supporter — Musk still controls and runs the company. 

If we go back to Musk’s original stated reason for acquiring Twitter, which he reiterated shortly after he first took control of the company, he said it’s “important to the future of civilization to have a common digital town square, where a wide range of beliefs can be debated in a healthy manner, without resorting to violence,” and that “there is currently great danger that social media will splinter into far right wing and far left wing echo chambers that generate more hate and divide our society.” 

Musk went on to critique mainstream media, “in relentless pursuit of clicks, much of traditional media has fueled and catered to those polarized extremes, as they believe that is what brings in money, but in doing so, the opportunity for dialogue is lost.”

But now, Musk is shifting Twitter toward the polarized, echo-chamber model of media he criticized. It’s not just that he is welcoming more right-wing voices onto Twitter. He’s also enabling them to seem more authoritative, boosting their voices, and allowing more controversial speech. 

Musk has allowed neo-Nazis, white supremacists, and other hateful accounts back on his platform under his “freedom of speech but not freedom of reach” approach to content moderation. While Musk says that Twitter doesn’t endorse hateful content and won’t amplify it in people’s feeds, the fact that these accounts are allowed on the platform has turned off some users. But conservative accounts have embraced Musk’s new hate speech policies, and paid their way into being verified under Musk’s new check mark system, which allows their content to show up higher in replies and comments. At the same time, Musk has stripped check marks from many media organizations, reporters, and politicians who refuse to pay for verification, making Twitter’s new check-marked class look markedly more right-wing. And Musk has repeatedly broken his promises to allow speech he disagrees with onto the platform by temporarily banning journalists, comedians, and others who draw his ire. 

DeSantis, like Musk, has been accused of contradicting his free-speech values: The governor has endorsed controversial new legal reforms in his state restricting how schools can talk to students about gender and race in the classroom.

Musk could embrace the new reality that Twitter is now a place where conservative voices are often welcomed by the company’s owner, and liberal ones are attacked. But somehow, he continues to hold on to the digital town square dream despite his failure to bring that vision to fruition. 

In a Wall Street Journal talk on Tuesday, Musk said he “absolutely” wants to also interview Democrats and politicians across the spectrum. It’s unclear, though, whether Democratic politicians would want to go on Twitter to interview with Musk, who — like it or not — is now seen as being publicly aligned with conservatives.

“I am interested in X/Twitter being somewhat of a public town square where more and more organizations post content and make announcements on Twitter,” Musk said in the WSJ interview.

But what kind of public town square is Musk building if one side is welcomed with open arms and the other is attacked? The Atlantic’s Charlie Warzel has argued that Twitter has evolved into a far-right-wing platform. Sara Fischer and Mike Allen at Axios wrote that “the center of media gravity” is moving from Fox News to Twitter. One could also argue that it’s not quite there yet because of all the left-leaning or apolitical holdouts who, despite all the drama, just can’t seem to quit Twitter. But if it continues to alienate a wide swath of users, Twitter will turn into something more similar to Trump’s Truth Social or the now-defunct Parler: echo chambers of conservative voices.

Musk has the power to drive attention to the politicians he favors by controlling the fire hose of information on Twitter. Musk has the ability to boost DeSantis, Carlson, and other conservatives he’s partnering with on Twitter so that they’re plastered all over people’s timelines. He has already given these figures free promotion and a launching pad at pivotal stages in their careers. Ultimately, it’s Musk who controls the Twitter fire hose of information. Let’s also not forget that he tweets to his 140 million Twitter followers on a daily basis. 

A number of outlets, including this one, have recently reported that Twitter as we know it is dying. But a more updated conclusion is that it’s being reborn as something else entirely, something certainly more right-leaning, and possibly even more of a polarized hellscape than the Twitter before it. In any case, it’s clear now that the Twitter Musk is building is not the all-inclusive digital square he promised, but it is the one he wants. 

Update, May 24, 7:50 pm ET: This story has been updated with details from the DeSantis Twitter broadcast event.

]]> Shirin Ghaffary <![CDATA[Elon Musk won’t stop tweeting his way into trouble]]> https://www.vox.com/technology/2023/5/20/23730607/elon-musk-conspiracy-twitter-texas-shooting-bellingcat-taylor-lorenz-psyops 2023-05-19T18:59:27-04:00 2023-05-20T07:00:00-04:00
Twitter owner Elon Musk at a meeting with the French Minister for the Economy and Finances in Versailles, France. | Ludovic Marin/AFP via Getty Images

Despite criticism for continuing to engage with conspiracy theories, Elon Musk is posting through it.

In the past five days, Musk has doubled down on a conspiracy theory about the Allen, Texas, shooter; tweeted that billionaire philanthropist George Soros (who has long been the target of antisemitic conspiracy theories online) wants to “erode the very fabric of civilization” and “hates humanity;” and promoted a quickly debunked rumor that falsely claimed Washington Post journalist Taylor Lorenz was related to the founder of the Internet Archive.

When CNBC reporter David Faber asked Musk about his promotion of conspiracy theories in a live interview on May 16, Musk acknowledged that he does support these theories, like the Hunter Biden laptop scandal, because “some of these conspiracy theories” like the Hunter Biden laptop story “have turned out to be true.” When Faber pressed Musk about whether his promotion of conspiracy theories is hurting Twitter’s reputation with advertisers, Musk was defiant.

“I’ll say what I want to say, and if the consequence of that is losing money, so be it,” said the billionaire.

Musk’s argument is that because conspiracy theories can sometimes be true, it’s okay to entertain them. Musk is correct that the Hunter Biden laptop story did turn out to be real, even though many in the media questioned its initial veracity. Twitter, the company Musk now owns, even controversially blocked the New York Post story breaking the news. It’s also true, as my colleague Zack Beauchamp has explained, that there can be valid, non-antisemitic reasons to criticize powerful figures like Soros.

But many of the conspiracy theories Musk has promoted to his 140 million Twitter followers have proven false. And some argue there’s potential danger in Musk spreading misinformation or heated rhetoric. The way Musk characterizes Soros, for example, stands to “embolden extremists,” according to Anti-Defamation League president Jonathan Greenblatt. Musk accused Greenblatt of defaming him and vehemently denied he’s personally antisemitic, saying “if anything” that he’s “pro-semite,” in the recent CNBC interview. 

Let’s clarify right out the gate that Musk has a First Amendment right to say what he wants, with some limited legal exceptions. And as the owner of Twitter, a private company, he has the final word on what speech is or isn’t allowed on Twitter. But because Musk is the leader of the platform and one of the richest, most influential people in the world, his actions have consequences, both tangible and symbolic. Moreover, Musk’s tweets serve as a hindrance to the very real challenges Twitter faces to win back more advertisers who have left the brand, in large part over concerns about Musk’s behavior.

A closer look at Musk’s “flirting” with conspiracy theories 

Musk’s engagement with conspiracy theories in recent weeks hasn’t been direct. He’s been described as “flirting” with these theories, oftentimes replying to a conspiracy theory tweet rather than directly tweeting it himself. He’s often taken a “just asking questions” approach, especially on subjects where there’s an information void.

For example, the week of the recent Texas shooting, early coverage of the incident, including that of open source intelligence research group Bellingcat, claimed that the suspected shooter was a white supremacist, based in part on his social media activity. Soon after, Musk jumped into the fray. He responded to a cartoon meme posted by an account called “The Redheaded libertarian” that raised questions about whether the suspected shooter, who is of Hispanic origin, could really be a neo-Nazi, and whether his account on a Russian social media app was legitimate.

“Didn’t the story come from @bellingcat, which literally specializes in psychological operations? I don’t want to hurt their feelings, but this is either the weirdest story ever or a very bad psyop!” Musk replied.

Bellingcat founder Eliot Higgins has long denied theories about his organization having ulterior motives, saying in a past interview that its contributors are just “people with laptops and free time.” But even if you don’t believe Higgins or Bellingcat on the subject of the Texas shooter, you don’t have to take their word for it. That’s because Texas law enforcement confirmed on May 9, the same day Musk posted his initial tweet casting doubt on the shooter’s white supremacist ties, that the suspected shooter had “Neo-nazi” ideation as confirmed by patches and tattoos on his body. 

Despite this additional evidence, Musk doubled down on his disbelief about the shooter’s ideology a week later in the CNBC interview, saying there is “no proof” that the shooter was a white supremacist. Either Musk doesn’t know about or is choosing not to mention law enforcement’s account here. That doesn’t mean there’s never reason to question public officials’ account of a crime, but Musk didn’t do that here; he just avoided acknowledging a notable piece of evidence that contradicts his theory.

Race and crime seem to be points of particular interest for Musk. Last week, he replied to a tweet that contained a chart claiming to show that Black-on-white crime is higher than that of other categories of crime by race. Many have criticized the chart for being skewed, but Musk seemed to endorse it. “Odd, why would the media misrepresent the real situation to such an extreme degree?” Musk tweeted.

Others have said that the chart Musk is referencing is misleading because it’s adjusted for percentage of the population: Because white people account for a larger percentage of the US population, it makes sense that they are victims of crime in larger numbers. Also, the chart didn’t include white-on-white crime, which is among the highest buckets of crime broken down by race in the US. 

As with many of Musk’s conspiracy-theory-flirting tweets, when put together, we start to see a fuller picture emerge: a specific grievance. In this case, it’s how the media portrays race and crime.

What’s going on with Twitter and Musk’s businesses

It’s notable that Musk is tweeting controversial tweets during a particularly critical moment for the future of Twitter.

The company has lost more than half of Twitter’s top 1,000 advertisers since Musk took over in November, according to January data from analytics company SensorTower. Some major companies, like Disney, Amazon, and Apple, have continued to spend money on the platform, and Musk has said other advertisers are coming back. But he’ll need more advertisers in his corner if he wants to turn Twitter’s business around.

Musk announced last week that former NBCUniversal advertising executive Linda Yaccarino will be replacing him as CEO. Given her street cred in the advertising industry, she’s seen as someone who can help repair Twitter’s reputation with major brands.

“If anyone can do it, it’s Linda,” one advertising executive, who spoke on the condition of anonymity, told me last week. It looks as though Yaccarino has already managed to help bolster Twitter’s image with GroupM, one of the top advertising agencies in the world. The agency, which is owned by WPP, had designated Twitter as a “high-risk” company for its clients to advertise with, but, according to the Financial Times, recently removed the designation.

If Musk’s tweets were gaining him positive attention or drawing attention away from some other, bigger issue, you could say his tweets serve as a useful distraction or a savvy way of gaining free media coverage. Instead, Musk’s behavior is self-destructive. His controversial tweets could exacerbate the most pressing problem with Twitter: its relationship with advertisers.

Musk’s continued conspiracy theory flirtation threatens to stymie any progress Twitter is making on that front. Back in November, Musk’s tweet promoting a conspiracy theory about the violent attack on Nancy Pelosi’s husband was seen as one of the turning points for many advertising industry insiders in their assessment about whether or not to continue advertising on Twitter.

It’s clear that Musk’s repeated promotion of conspiracy theories has real consequences for his business. If there’s a deeper reason for why Musk continues to promote these theories despite the consequences — beyond merely saying what he wants to say — is anyone’s guess. Whatever the reasoning, Musk is posting through the firestorm and getting all kinds of attention for it. Maybe that alone is worth the trouble. 

]]> Shirin Ghaffary <![CDATA[Instagram’s co-founder explains why he’s starting over]]> https://www.vox.com/technology/2023/5/19/23727598/kevin-systrom-instagram-artifact-new-app 2023-05-19T12:14:36-04:00 2023-05-19T06:00:00-04:00
Instagram co-founder Kevin Systrom has created a new app called Artifact with fellow Instagram co-founder Mike Kriefer. | Callaghan O’Hare/Bloomberg via Getty Images

Instagram co-founder Kevin Systrom thinks that the social media industry is ready for something new.

Major social media platforms have become better than ever at capturing our attention by optimizing their algorithms to entertain us with viral videos and funny memes. But while people are consuming more content, they’re often posting less, according to Systrom.

“People have flocked to services like TikTok or Twitter or Facebook less to connect with their friends … and more and more to be entertained,” Systrom said. 

He added, “I think people want value and entertainment, but they don’t want to be in the middle of a digital fistfight.” 

Systrom, along with fellow Instagram co-founder Mike Krieger, sold Instagram to Facebook for $1 billion in 2012 and continued working at the company until 2018. But now, Systrom thinks there’s an opening for a new kind of engaging app that can better inform the public. 

That’s why in January, Systrom and Krieger launched Artifact, a personalized social news reader — although it’s not just for news, Systrom says — that shows you high-quality written content, recommended to you based on AI. It’s been called a “TikTok for text” because, much like the popular video-sharing app, it’s designed to predict what to show you based on detailed insights about what you’re looking at, what your interests are, and what you’re clicking on.

Systrom hopes Artifact can help solve a major problem: how to help writers reach interested readers at a time when the online advertising industry is facing a slump, Facebook is backing away from news, and AI-generated content is threatening to upend the news industry and blur the line further between what’s fake and real. 

“IT feels like smart people should insert themselves into existential crises for the world — and hopefully we qualify as relatively intelligent, having done something in this world before on social,” said Systrom.

Part of Systrom’s plan, he says, could be to eventually allow independent publishers to post on the platform, rather than just major media organizations.

“The internet is this wonderful place where potentially anyone can be a publisher just like anyone can be a creator on TikTok or a photographer on Instagram,” Systrom said. “And it feels like that opportunity is untapped.”

But the business of written words — especially news articles — is a notoriously difficult one compared to, say, funny meme videos. Artifact will face tough competition from apps like TikTok to hold our attention. Newspaper articles and magazine features “may not make you laugh as much” as video, Systrom said, but the text medium is “enormous in terms of its effect on society.” That’s why he’s trying to figure out a way to make it work.

Systrom and his team will also have to figure out how to avoid repeating the mistakes of social media’s past when it comes to letting harmful or misleading content go unchecked. Artifact has a content moderation policy that bans things like hate speech. The company says that isn’t too much of a problem so far because the app hand-selects which publishers are shared on the platform, and they select for high quality. But Artifact recently started allowing comments, which opens it up for more content moderation problems. One way to mitigate that is the app assigns users Reddit-inspired “reputation scores” based in part on how much other users upvote someone’s comments.

In a far-ranging interview, Systrom reflected on his time working in the social media industry, saying that one of the things he was “always questioning” was whether he was “providing value to people.” Now, Systrom is hoping that Artifact will help bring meaningful articles to people’s attention, whether that’s from independent writers or major publications. 

“When people engage with the service, are they learning something new? Are they being more informed?” Systrom said. “And the second time around … hopefully, by us doing our job right, people will learn more about the world, and they’ll be more informed citizens.”

The following conversation has been edited for clarity and length.

You were one of the pioneers of early social media. What do you make of the social media landscape right now? 

I think that there are two forces. 

On the corporate side, companies have gotten very good at figuring out how to get your time. They’ve been optimized by engagement algorithms. Basically the single goal is: how do you spend more time on the service? And unsurprisingly, that has led to the type of content that optimizes the system being perhaps more entertaining, more inflammatory, and basically just more engaging. And I think what you see is people have flocked to services like TikTok or Twitter or Facebook less to connect with their friends and more and more to be entertained. Basically, all of these things have moved to one giant entertainment network. 

And on the other side, I think consumers are starting to raise their eyebrows, because I think they want two things, I think they want value, and I think they want stability. The value is: I’m spending all this time, what am I getting? Do I get a laugh? Am I learning something new about the world? Am I seeing something I wouldn’t have otherwise? And the stability part is interesting, probably more so in the last five years or so. 

Kevin Systrom and Mike Krieger won a Webby for “Breakout App of the Year” in May 2012, just a month after Facebook bought Instagram for $1 billion.

I think people want value and entertainment, but they don’t want to be in the middle of a digital fistfight. They don’t want to be the target of someone else’s unbounded anger, or they don’t want to put themselves out there, take a risk to try to make a funny video, and get put down. And to me, the side effects of that are that it’s largely driven people into “consumer mode” rather than “producer mode.” So people are consuming vast amounts but are producing far less than they used to. They want the value but they don’t want to necessarily interact with a bunch of other people. They don’t want to put themselves out there and be taken down with a bunch of replies and tweets. And by the way, the people that are immune to this — like if you think of this as a system, if you start out with everyone just kind of like producing, the people that are immune to other people’s feedback and other people’s attacks, they’re usually not the people you want producing content because they’re the ones that will produce with impunity and chat with impunity. And I think that leads to a certain type of content dominating these networks. 

It also means, on the consumption side, the algorithms are fairly focused on driving time spent. People who get really good at optimizing videos for time spent are the ones that get the most distribution, and those are the creators. You have to be a professional to be that good. You can’t just be anybody. 

So anyway, wrapping up those two forces — the drive toward extreme engagement optimization, and on the consumer side, wanting more value and wanting stability while you extract that value — leads there to be a bunch of people who are trying to do new and interesting things. 

Also, just like the context around this is you have a very large company, Meta, who has been extremely successful dealing with disruptions on the outside. You have an international player in ByteDance, trying to figure out if they can succeed in the United States and maintain a business in the United States long term. You also have Twitter, which is now run by a very different person than it used to be run by with very different values and ways of working. 

So I can’t tell you exactly what will happen. But I guess my rule of thumb in business is out of turbulence, out of disruption, new things rise. And I think that AI kind of coming into its own, rather than being just this buzzword we all use — that turbulence plus AI feels like a bunch of really interesting things are about to happen. That’s my long answer.

Can you talk a little bit more about why people are burned out these days on Instagram or Facebook or some of the older social media platforms? Have these platforms gone too far?

I’m not sure they’ve gone too far. I think they’ve just become too successful. You know, if you hang out in a small group of people, you’re likely to be much more talkative and much more risk-taking: You’ll show someone your photos, you’ll talk to them about what you’re doing. The second you grow your audience to a certain size, it becomes difficult to be yourself. 

I remember there was this moment at Instagram where we had lots of college-bound high schoolers basically either making their accounts private or going under pseudonyms because they were worried about college admissions teams being able to see them, people worried about their bosses seeing them on Instagram. 

“I think people want value and entertainment, but they don’t want to be in the middle of a digital fistfight”

What makes you successful is also the thing that will end up being your downfall. Which is that you’re successful and now everyone’s on the app, and anyone can see what you’re doing. And you need to manage that growth over time. It’s not because Facebook and Instagram aren’t wonderful products. They are. It’s simply because they are so large now that you are on display, and it’s very hard to feel like you can be yourself when you’re on display to over a billion people. 

I think that’s part of why there’s this natural cycle in startups that attack social problems: People always want the next smallest thing. It’s this never-ending cycle of people just wanting smaller spaces to be themselves. As a large network, you don’t have a choice but to focus on influencers and businesses, because those are the only people and accounts that are willing to continue posting at that level. They aren’t interested in posting on you when you’re small, but they are increasingly interested in posting on you as you get bigger because it provides business value. But it does mean the community changes a lot. And it means the user experience changes a lot. And you just have to be mindful of that as you grow.

I do think a lot of people feel like they don’t want to post on Facebook or Instagram because there’s this expectation that it has to be really professional. Or because, like you’re saying, there’s too much fighting going on, and they don’t want to get in the crosshairs. But do you think that TikTok challenged that a little bit?

What I love about what ByteDance did with TikTok was that they made it what I call a content meritocracy. Content’s great? It will get a lot of distribution. And that’s the first time, I think, in social media that that had been true. In the past it was if, in general, you can build a following — whether that’s through how famous you are or how great you do at marketing yourself — you will get distribution for your content, period. And sure, plus or minus, the feed ranking will do its job. But generally speaking, the only way to be big is to have been big already, or to grow. TikTok kind of upended that and said, “Well, you know, it matters less about the person who produced the content. It’s the content that actually matters.” 

So what we’re going to do is millions of small little tests with content to see what sticks, and then we’re gonna promote whatever sticks to the next level, and then promote whatever sticks to the next next level, and so on and so forth. And what’s cool about that is you get this meritocracy, where if you create great content, you will be distributed, and I believe that actually exists not just in videos, but it exists in written content, whether that’s blogs or articles in a large newspaper publisher. Basically, I think content should win, meaning great content should be distributed, regardless of how well-known or big the producer is. And I think that, in and of itself, is an opportunity across many different mediums.

News is a tough nut to crack. It’s a notoriously difficult business from a revenue perspective. And then it’s also difficult in terms of the political baggage that can come with it. So with Artifact, why venture into this unwieldy territory rather than just making a fun video or meme-sharing app?

Well, first, I’ll just challenge the notion that we’re just in news. A lot of written content on the web I wouldn’t characterize as news so much as information.

Some of my favorite blogs to follow are modernist architecture blogs, and they have nothing to do with news. It’s just beautiful photos and really interesting design. I love to cook, and I would not consider any recipe news per se. Let’s see, what else do I love? I love travel — at least the idea of travel. I feel like I’ve been working too hard to travel. But the idea of travel is really neat to me, and I live vicariously through all these amazing travel bloggers and folks that produce, you know, top 10 lists of hotels around the world or whatever.

So there’s a lot out there that doesn’t qualify as news, but I will agree it qualifies as written information — or at least articles, more broadly. It is true that in general, video content can be more engaging and more viral. And it’s also harder to produce. It’s not clear to me if it’s as informative. 

Screenshots of the news app Artifact

For instance, I obviously used to live in the world of images and video. And one of the things I always found lacking was, am I providing value to people? Like when people engage with the service, are they learning something new? Are they being more informed? And so the second time around — hopefully by us doing our job right — people will learn more about the world, and they’ll be more informed citizens. 

It feels like smart people should insert themselves into existential crises for the world. And hopefully we qualify as relatively intelligent having done something in this world before on social. So really, it comes down to the fact that I think there’s an untapped opportunity in lots of content out there. And then secondly, I think it provides more value to the user if we do our jobs right. I’m not claiming that text itself can be as entertaining as a viral video, as you described. But in general, that doesn’t mean that it will be a bad business. It just means that there’s different characteristics and can be potentially enormous in terms of its effect on society. It may not make you laugh as much. But I think it could be really important.

I have to agree with you, otherwise I wouldn’t be in this extremely stressful and relatively low-paying business. 

I also think that we should have a world where people do investigate, people do question, people do publish. And right now that is relatively limited. It’s limited to people with jobs at places that can provide distribution. And to me, in the long run, I think that we need to find more ways of just anyone producing. 

I think Substack is super interesting because it allows people to have a platform to self-publish. And I think that we will always have large established publishers because the world needs those for the editorial oversight, the resources to do very expensive investigative journalism, etc. But also it feels to me like the internet is this wonderful place where potentially anyone can be a publisher, just like anyone can be a creator on TikTok or photographer on Instagram. And it feels like that opportunity is untapped. I’m not saying we are attacking that opportunity currently, but it certainly seems that that’s the direction that the world is heading.

You bring up Substack. Who do you see as your competitors in this space? Is it Substack? Is it Twitter? Is it TikTok? 

It’s interesting, people ask if we compete with Substack. I actually hope to be a wonderful partner to Substack. They have these amazing writers who want distribution, and we have this user base that wants content, and we have the targeting to be able to help relatively undiscovered writers find their audience. And we’re still fairly small. I mean, we started a few months ago, but eventually, the idea is that we can be the distribution for anyone that wants to self-publish, no matter what platform you choose, whether it’s Substack or one of the others. I think that’s the goal long term. So I don’t see us competing directly with any of those people. If anything, I hope to be a great partner. 

I do believe where you discover content, generally, includes the companies that I think compete for the use case that we provide. Of course, there are many apps that are like Artifact. There’s Apple News, there’s SmartNews, there’s NewsBreak. There’s lots of stuff out there that will feed you yet another local story of, I don’t know, a car crash or a fire. A lot of the content tends to be fairly general and fairly clickbaity on a lot of these platforms. And I think what we’re trying to do is provide a very high-quality experience for the people who use us, which tend to be very tech-driven, AI-interested users.

“You don’t need to build in a massively viral sharing loop where everyone spams their contacts and everything”

We ask people to select interests when they sign up. AI was selected by some crazy percentage of people that signed up. And I think that’s because we talked a lot about how we use artificial intelligence to help with matching content and people. And as a side effect, I think what has happened is the data that the algorithm has been trained on is particularly good at the tech stuff. And we’re doubling down on that because we see that as a particular opportunity — versus trying to be good at yet another service that feeds you the headlines in politics today, which I think you can get anywhere or better from some of the major publishers, honestly.

Artifact has been described as like a TikTok for news, or maybe you would say for articles. How is it TikTok-like with regard to recommending content based on using AI?

There are two things. One is you try to model people’s interests. You try to figure out what someone is interested in. They are interested in space exploration. They’re interested in cryptocurrencies. They’re interested in a very specific type of Android. That allows you to narrow down the enormous corpus of data that we get every day to a subset of articles that may or may not be relevant to that person. But then you have the problem of saying, “Well, okay, we know these are relevant, but are they good? And what’s cool about this?” I think similar to TikTok is just keeping track of whether or not someone else has engaged with that article in the past allows you to very quickly come to a conclusion of whether or not something is engaging or not. 

You’ve recently added some features that could help the writers of this content better understand their audience. What value does Artifact bring to writers, or is Artifact maybe going to pay writers one day?

Are we paying anyone yet? No, we don’t make any money, and we don’t pay anyone yet. But I can absolutely imagine a future where writers use this to understand their audience and connect with their audience more clearly or more efficiently. And eventually there will likely be a place for independent publishing as well. Although we’re so small right now, that doesn’t really make much sense. And right now, I basically think that we’re trying to provide value and just understanding what’s working, what’s not. That’s what we’re focused on.

There are a lot of journalists who are fed up with Twitter right now. Do you think that this presents an opportunity for Artifact? 

I assume that you use Twitter not for distributing links to your articles and telling people to go read your new article on X, Y, or Z, but rather to stay in the know, one, and two, to have interesting back-and-forth conversations with people about the topics of the day. 

And, yeah, Artifact has some of those. But we’re too small to have that conversation at large. In fact, Post News and BlueSky are as well. All of these things are too small to really have that critical mass right now. And I think it remains to be seen if any of them will also do a good job at that job of having the conversation of the day with other like-minded people.

Kevin Systrom speaks with Andrew Ross Sorkin of the New York Times at the DealBook conference in 2019.

But in terms of discovering what’s happening in real time, that might actually be a lot easier to compete with. We’ll have to see if we’re good enough. But I believe there’s an opening there. I think there’s less of an opening in terms of having conversations with people in your industry only because, as I’ve seen it, social networks are just really sticky. I mean, name the number of controversies that all these big companies have gone through — whether it’s Facebook or Snap or anyone — people will say they’re shuttering their account. And then they’re back the next week. 

I think history would tell us that these networks are a lot stickier than you’d imagine. So if you’re going to do something to compete in this world, you need to do something fairly different to win. And that’s part of why we’re starting with the article-first version of this and the extreme personalization version of this. Because we believe that is fundamentally different than how you consume content on these other networks.

What do you call Artifact? A social media app? Or are you not putting it in that bucket?

Well, we’ve got comments now, which is pretty exciting. I mean, it’s small because we launched — what, it was like a week and a half ago, or two — it’s all a blur at this point. But it’s definitely social in that respect. And one of the things I didn’t mention that I should have mentioned is having comments does differentiate us in the long term. Because we clearly have a social angle on the news and written content that I care deeply about, but I think a lot of these larger apps do not.

Is it hard to grow an app virally if you’re just catering to one very specific group, like tech enthusiasts? I guess if anyone knows a thing or two about making a viral app based on people sharing stuff they’re passionate about, it’s you, the co-founder of Instagram. 

I do have the humility to know this is hard, and not show up saying that I figured it all out. But what I’ll tell you is, we rewrite the history of every one of these apps as if they were just a sensation from the beginning. All of these apps had some amount of growth, but they all started — by “all” I mean every social media app, including Instagram, Facebook, and Twitter — they all started with a very specific type of person who used them initially. And only after years and years and years did they see expansion. TikTok might be the only example that grew within a few years rather than years and years and years. Nonetheless, they all had groups of people — reference groups, I’ll call them. For Instagram, it was definitely photographers and artists in the Bay Area. And it was not the app it is today. 

There’s the business school book, Geoffrey Moore’s Crossing the Chasm. I’m not sure if you’ve ever read it, but it talks about beachheads, and the whole idea is you have to start specific to then get broad. But if you start broad from the beginning, you never capture enough interest. And that’s how we’re thinking about growing.

So it sounds like you’ve learned some lessons from the success of Instagram. Are there also any regrets from your time at Instagram? Any lessons you’ve learned and don’t want to do again?

I would say that we’re so small, and it’s so different. I think the lessons learned would be applicable if we get to be hundreds of millions of users. At this stage, you know, we were 13 employees when we were acquired by Facebook, and we’re seven now at Artifact. So we’re in the stage where we were so early that most of the lessons I learned from going through Instagram aren’t applicable yet. 

That being said, I will say the one thing we are laser-focused on is just doing our job well. That means when you open up the app, do you get a great experience? Because I believe if you do that, the growth takes care of itself. 

You don’t need to build in a massively viral sharing loop where everyone spams their contacts and everything. Yes, that can be helpful. But is that the type of app I want to build? Absolutely not. What I care about is providing value. So if it does the job well — and that is hard because personalization is not an easy science — then things will take care of themselves. So we’re just focused on being a great app and people loving it. If people love it, then good things will happen.

A version of this story was also published in the Vox technology newsletter. Sign up here so you don’t miss the next one!

]]>