Hello! Next Thursday, April 24, I’ll be moderating two fireside chats—with Runway cofounder Alejandro Matamala Ortiz and F-35 pilot Justin “Hasard” Lee—at Artist and the Machine’s AI & Creativity Summit in Brooklyn. The event promises to be an invigorating exploration of the intersection of technology and art from multiple perspectives, and I hope to see some of you there.
Sam Altman wants to build a social network. Given the OpenAI CEO’s unbridled ambition—and the potential to turn 400 million ChatGPT users into some semblance of a community—it would be weird if he didn’t. And the timing makes sense: On Tuesday, The Verge’s Kylie Robison and Alex Heath reported that OpenAI is looking to capitalize on the new ChatGPT image-generating features that have already been going viral on other networks since arriving late last month. Whether the company plans to fold social aspects into ChatGPT itself or create a new app remains to be seen.
The news of Altman’s interest in bootstrapping an OpenAI social network around ChatGPT came less than three weeks after Elon Musk announced that his AI company, xAI, had acquired his social network, Twitter X for $45 billion. It’s still unclear whether this move amounts to more than the world’s richest man shuffling assets around the way you or I might rearrange our shelf of Funko figures; after all, xAI’s Grok chatbot already occupied prime real estate inside the X app. But xAI, whose wellspring of AI technology and talent is, by all accounts, formidable, now owns a decent-size (if reputationally challenged) online community. That gives it access to hundreds of millions of potential customers, plus a vast, endlessly replenished stream of content it can use to train its algorithms. The potential to do interesting things is there, if Musk can divert his attention from dismantling civic institutions for a moment or two.
Even if it’s obvious why OpenAI and xAI might want to meld their respective generative AI engines with social platforms, it won’t be easy. Consider what’s going on at Meta, which has more social-network users than any other company and owns a top-tier LLM, Llama. So far, AI’s impact on Facebook has been to junk up feeds with spam posts about imaginary people and pointless interjections from the Meta AI bot. If there’s a way for the technology to make itself welcome in a communal setting, it probably doesn’t look anything like this.
For all the ways generative AI is astounding—it recently coded my dream app for me—social networking may be one of the tougher assignments for it to crack productively. The technology excels at churning out lowest-common-denominator content, but online gathering places need less of that, not more. Nor does its uncanny glibness mean it’s ready to join conversations the way humans do. Using machine vision, for example, the Meta AI bot can suss out the gist of a posted image, but that has little to do with whether it can say anything interesting about it. In my encounters, the conversation starters it’s generated have been stultifyingly synthetic.
Still, when I heard about OpenAI’s social aspirations, my instinct was to be intrigued rather than repelled. For one thing, many of us have spent the last few years obsessing over the company’s tools and sharing our creations: I started tweeting DALL-E 2 oddities in August 2022, months before there was a ChatGPT. The company should be able to facilitate that discussion by building some community infrastructure of its own, such as ways to post items for public consumption and comment. I’ve been disappointed by its failure to do much with the store for custom GPT applets it launched in January 2024, which could have been a springboard for features that let ChatGPT users talk to each other. The vitality of its newest image-generation technology is an even better such opportunity.
More socially aware AI products might also help counteract the AI’s tendency to suck users down rabbit holes of solitude. Recent research conducted by OpenAI itself suggests that heavy ChatGPT users tend to be lonely, though it’s unclear whether that’s because the chatbot fosters loneliness or simply a sign that lonely people are drawn to it. Either way, I find my own jags with AI chatbots to be both addictive and isolating in a way that doesn’t feel entirely healthy—at least when I discover I’m still at it an hour or two after I meant to go to bed. Anything that nudged AI devotees back toward engaging with society couldn’t hurt.
Part of my measured willingness to believe OpenAI could construct a worthwhile social network—maybe, in theory, if we’re lucky—stems from the fact that the company has shown it can create software that’s pleasant to use. Any community it built would be a fresh start, which is not true of xAI/X and Meta, both of which are cobbling together AI experiences atop social platforms that are well into middle age, and showing it. That guarantees nothing: Google+ was also quite pleasant but couldn’t overcome Facebook’s deeply entrenched place in people’s lives, particularly at that time. But an OpenAI social network wouldn’t have to displace Facebook to have value. In fact, the less Facebook-like it was, the better the argument for it existing at all.
If OpenAI’s project turns out to be a soon-abandoned lark rather than a top priority, somebody else might give AI a measure of social grace. For instance, Microsoft consumer AI chief Mustafa Suleyman recently told me that the company is working on priming its Copilot companion to participate in conversations with multiple humans at once, with a sensitivity to their varying interests and attitudes. Or maybe a company that doesn’t even exist yet will do the job—which would make sense, since every social network that has ever mattered has been the brainchild of a tiny startup.
The one scenario that seems implausible is that tomorrow’s social experiences won’t have a far heavier element of generative AI than anything that exists today. We already know what can go wrong when they get smooshed together. Fingers crossed we’ll get to see what can go right.
You’ve been reading Plugged In, Fast Company‘s weekly tech newsletter from me, global technology editor Harry McCracken. If a friend or colleague forwarded this edition to you—or if you’re reading it on FastCompany.com—you can check out previous issues and sign up to get it yourself every Friday morning. I love hearing from you: Ping me at [email protected] with your feedback and ideas for future newsletters. I’m also on Bluesky, Mastodon, and Threads, and you can follow Plugged In on Flipboard.
More top tech stories from Fast Company
Facebook Groups are fueling a black market for Uber and DoorDash accounts, says a new reportA TTP report finds Meta’s group features are enabling fraud at scale—allowing unvetted drivers to bypass safety protocols on Uber and DoorDash. Read More →
‘I would get way more views if I didn’t help thousands of people’: MrBeast defends his philanthropy‑as‑content strategyThe world’s biggest YouTuber posted stats showing his charity videos rank lowest among recent uploads. Read More →
How to clean up Google Photos to save space—and moneyOne of the top photo sites starts free but can get expensive as your collection of pictures and videos balloons. Read More →
GM just made an EV Corvette concept car. Could it be a new American icon?The Corvette was once a symbol of American progress and industrial might. In the EV age, the car is ripe for reinvention. Read More →
TikTok starts testing Footnotes, a new feature that looks a lot like X’s Community NotesTikTok said it will open access to start contributing to Footnotes in the coming months. Read More →
This smart Reddit answer engine puts Google Search to shameReady to get crowdsourced intelligence with next to no effort? Read More →
No comments