Random Image Display on Page Reload

Hey Zuck, Get Those Robots out of My Social Feed

Sep 29, 2023 9:00 AM

Hey Zuck, Get Those Robots out of My Social Feed

By injecting chatbots into Meta’s social platforms, Mark Zuckerberg threatens to undermine his company’s original mission: to connect humans with other humans.

Mark Zuckerberg speaking on stage with various celebrity AI avatars on a screen behind him

Meta CEO Mark Zuckerberg announced new AI chatbots modeled on celebrities at a company event this week.Photograph: David Paul Morris/Bloomberg/Getty Images

Mark Zuckerberg’s keynote presentation at the Meta Connect event this week started late. The delay was on-brand for the company’s decade-long project to make virtual reality mainstream. Back in 2014, a demo of the then-primitive Oculus VR headset hit Zuckerberg like a lightning bolt, and within weeks he owned the company. He started talking about how a digital version of reality was going to be the next computing paradigm in about, oh, five or 10 years. More than nine years since that first demo, we’re still waiting. For a time there was a lot of excitement about an impending Metaverse, but the buzz is now barely audible. These days everyone is enchanted by all-knowing AI chatbots, which are making their claim to be not just the future of computers but the future of everything. What’s a paradigm chaser to do?

The answer came in Zuckerberg’s bifurcated presentation at the Meta Connect event this week. Yes, he still believes in mixed reality. He started by formally announcing the Meta Quest 3, a $500 headset that is actually better than the $1,500 “Pro” helmet the company recently abandoned, not least because it can also provide a really good augmented reality experience. Zuckerberg promised a profusion of digital objects that will become layered into the physical world, for things like games and fitness programs—so many that before long every room we enter will have more holograms than physical objects. After that bold prediction, he made an awkward pivot to Meta’s AI efforts, turning from a technology that has left many consumers indifferent to one that no one can get enough of. This was the moment Zuckerberg unveiled his bold new AI strategy for the era of ChatGPT. The gist of it is to use Meta’s advanced large language models to create chatbots injected into it various social platforms, including Facebook, Instagram, WhatsApp, and Messenger.

I have serious questions about this path. The original point of Facebook, a mission that Zuckerberg has at no time repudiated even after his company’s name change, is connecting people. You know, humans. For years whenever Zuckerberg talked about VR he would emphasize that it was going to be a social medium. VR in his view was all about hangouts, meetings, and getting you-are-there experiences from afar, largely anchored by Meta’s Horizon Worlds app.

Recently that messaging has shifted. A couple of weeks ago I got a demo of the Quest 3. A bunch of product managers and execs talked for over an hour about the headset and the new applications. But there was not a single mention of any improved social experience. One might have expected at the least a ground-up makeover of Horizon Worlds, since reports coming out of Meta have shown executives complaining that even employees ordered to use the app have been avoiding it. When I asked about the omission, the executives hemmed and hawed. In this week’s keynote, Horizon Worlds and social experiences hardly got a mention.

Zuckerberg’s AI presentation this week departed even further from his original mission. Meta is deep into the race to create super smart chatbots, and the big idea is to put them into social feeds. Some will appear as avatars, posting stories just like humans do. You will also find that a new contact has elbowed in among your friends and others you may follow, in the form of a chatbot called Meta AI that answers questions like ChatGPT does. This electronic interloper can even join in on your group chats.

Most Popular

Meta AI chatbot social app implementation demo

Meta AI, the company’s new chatbot assistant, can be accessed via several of the company’s platforms.

Meta via WIRED Staff

Zuckerberg took special pride in a “fun” use of AI—a slate of about 30 chatbots modeled on celebrities including Tom Brady, Kendall Jenner, and Mr. Beast. The stars somehow participated in the process of making the bots, but these are not celebrity stand-ins that pretend to be the real thing. That would be only vaguely unsatisfying. Instead, the Meta team assigned stereotyped personas—jock, fashionista, chef, and more—to the celebrities, who act out the roles in bot form via their voices and animated facial expressions. When you see Naomi Osaka on your feed, for instance, you’ll be getting even less of her than when you watch her hawk a product in a TV commercial. A Meta spokesperson later confirmed that there is no way that the Snoop Dogg avatar would ever make a reference to his choice of recreational substances, which you are kinda guaranteed every time he appears elsewhere.

One of the interesting early promises of social networks was that, among your missives from friends and families, you might find celebrities showing their real selves. When Taylor Swift uses Instagram to communicate with Swifties, that’s her, or at least a human posting in her stead. But Meta is now unabashedly pushing ersatz dupes of celebrities. Where’s the human connection in that?

The new Meta AI chatbot doesn’t pretend to be a person, but it too could undermine the original point of Facebook. Meta executive Li-Chen Miller showed us how she used the company’s new Smart Glasses to capture a video of her cat on an oversized hamster wheel and made the familiar lamentation that she was hard up for a bon mot to spice up the share. Meta AI provided a pretty good one: “Adobo the cat: where exercise meets existential crisis.” Yuks all around, right? Yes, until you think about it. We’ve reached a Cyrano moment in human communication. Isn’t the warming feeling of chatter among friends dependent on the chatter … actually coming from your friends? In this case both the creator and the recipient are reduced to spectators as the AI struts its stuff. This is the opposite of socialinteraction.

Most Popular

When I bring up these concerns to Meta’s VP of generative AI, Ahmad Al-Dahle, post-keynote, he accuses me of having “a dystopian point of view.” Adding AI bots to the various feeds will spur human connection, he counters. “I think these AIs are entertaining and can help people learn new skills that help them better connect with others,” he says, “supercharging your own capabilities to build better EQ and connect with people in more meaningful ways.”

Facebook was supposed to be about friends and family. It also urged us to expand our connections into a web of friends, colleagues, acquaintances, and friends of friends that its founder called “the social graph.” The world would be better, Zuckerberg promised, when humans bonded via his amazing social tool. But when did robots qualify for my social graph? Social feeds are zero sum. Every time I get diverted by some chatbot interaction, whether an automated replica of a human or a bot doing the work of a search engine, that’s one less chance to see a post from a cousin and respond. Even worse, one day I might wind up interacting with my cousin’s avatar, which might be wittier than he is but isn’t the flesh-and-blood person I’m interested in. And by the way, Mark Zuckerberg—if frigging virtual Tom Brady ever winds up on this Eagles fan’s feed, I will abandon your platform faster than you can deflate a football.

I have previously argued that Zuckerberg should split Meta in two. Give some other poor soul all the mishegoss of handling its wildly profitable but troubled social networks, and create a new and unburdened company with Meta’s formidable strengths in mixed reality and AI. This year’s keynote has only strengthened my conviction. The Quest 3 is an important leap for a number of reasons, not least of which is that Meta has been so far been outflanked in mixed-reality headset buzz by Apple’s flashy but super-expensive vaporware, Vision Pro. Following Zuckerberg’s presentation, Meta CTO Andrew “Boz” Bosworth dove into the details, dumping on Apple’s $3,500 headset (without mentioning it by name) with constant references to the Quest 3’s $500 price tag and abundant supply. (Apple reportedly has cut estimates of its first-year sales to less than a million units). “If you played a drinking game, and your keyword was mainstream and mass market, you’d be in trouble now,” Boz said. Though $500 isn’t exactly pocket change, this headset may indeed sell well and advance the vision that made Facebook change its name to Meta.

As for that overall vision of a mixed-reality paradigm shift where the line between digital and physical blurs to become imperceptible? That’s still at least 10 years out, and it may be even longer. But as Xerox Parc pioneer Alan Kay once said, we tend to overestimate the impact new tech will have in the short term and underestimate its future effects in the long run. I remember meeting with AI scientist Kai-Fu Lee, then at Apple, in the early 1990s, capturing our conversations with a cassette tape recorder. In five years, he told me, you won’t have to transcribe these talks or hire someone to do it. Five years later, when he was at Microsoft, Lee told me the same thing—wait five years. A half decade later, same story.

Most Popular

These days I record hundreds of interviews each year and nearly every word is transcribed accurately by an app whose annual cost is less than what I used to pay a human to render a single conversation into text. What was perpetually five years out was suddenly here and now. Same with AI models that talk like humans, originally envisioned almost 70 years ago, and since last November, unavoidable. That phenomenon might happen with the Metaverse as well in … let’s say 10 years? But I have no idea what that has to do with Facebook’s original mission.


Image may contain Label Text Symbol and Sign

In my book, Facebook: The Inside Story, I describe how Zuckerberg’s infatuation with virtual reality began with his snap decision to buy Oculus. It was all rooted in Facebook’s scary experience grappling to pivot from web to mobile. Virtual reality, he believed, would demand a similar paradigm shift. But even then, Zuckerberg saw the shift as a way to continue the company’s social mission.

On January 23, 2014, [Oculus CEO Brendan] Iribe and a small team flew up to Facebook. Since Zuckerberg’s glass-walled conference room was exposed (Zuckerberg found it a pain to draw down the shades that could provide privacy), they set up in Sandberg’s office. Zuckerberg put on the headset and began exploring a strange landscape with critters running around. One part of the demo particularly impressed Zuckerberg. It depicted a villa in Tuscany, Italy, and allowed the user to roam around, exposing beautiful vistas of the countryside. This is really cool, Zuckerberg thought. I’m clearly not in Italy—I’m in Sheryl’s conference room. But I really feel like I’m in Italy because everything I’m seeing makes me feel I’m there!

The next day, Zuckerberg emailed Iribe. “I was a little dizzy after taking off the headset,” he wrote, “but it’s clear where it’s all heading, and it’s amazing.” He wasn’t offering to buy Oculus yet. But five days later, he flew to Irvine himself for a more elaborate demo.

Most Popular

The second demo clinched it. In the space of a few days, Zuckerberg had concluded that virtual reality was not merely a cool potential feature, but something way larger. It was the next platform. Missing this would be like missing out on mobile. Zuckerberg was only two years removed from what he had considered a near-death experience when Facebook almost screwed up that pivot. Virtual reality, he figured, might be ten years out, but here was a company that was building the foundation. If Facebook owned it, and poured in money to make it happen, Zuckerberg not only would be ready for the next big paradigm shift. He would own it.


Image may contain Symbol

Hitesh asks, “How far away are we from witnessing a 100 percent AI-controlled society?”

Thanks for the question, Hitesh. Or maybe your bot asked it. Anyway, it’s a good one. The fact is right now a lot more in society is controlled by AI than lots of people realize. Many of the systems that keep things flowing are more complicated that humans can master, and we have gradually turned those over to AI. Face it, human beings aren’t doing a great job with the Texas electric grid.

You seem rightfully concerned about the issue of control. Obviously, when the people in charge of these systems employ AI to operate them, the intention is not to have robots making decisions that overrule the goals and principles that human creators intended. But pulling the plug is difficult. That’s not because the AI will intentionally shut us out like Hal 9000 in 2001. The problem is that when relying on AI to do what is beyond human capability, shutting them off means those systems probably won’t be able to run anymore. If you build fighter jets so fast and full of gadgets that humans can’t fly them, they are useless without AI.

Most Popular

The risks of this are obvious. Again, not that a Skynet-like global intelligence will organize the bots against us. But any computer system can fail or be corrupted or taken over.

AI is not going to control 100 percent of society, ever—if it did you might argue that society is over. But AI systems will control plenty more than they already do. That’s not necessarily bad, and it will enable efficiencies that benefit everyone. But unless we humans provide constant vigilance to maintain those systems, the potential catastrophes are almost as big as that climate thing that we also screwed up on.

You can submit questions tomail@wired.com. Write ASK LEVY in the subject line.


Hey Zuck Get Those Robots out of My Social Feed

How do you follow the hottest July and August ever? With the hottest September!


Image may contain Label Text Symbol and Sign

Bad enough that AI can beat us in Chess, Go, and Instagram captions. But mastering the game Diplomacy? Too much!

“We want to map it all,” says an archeologist of the ancient world. Isn’t that what Google said?

Most Popular

Your complete guide to the Sam Bankman-Fried trial.

The GOP lawmakers threatening to shut down the US government are click-seeking online addicts. In this case, bots could be an improvement.


Image may contain Logo Symbol Trademark Text and Label

If you buy something using links in our stories, we may earn a commission. This helps support our journalism.Learn more.

Get More From WIRED

Steven Levy covers the gamut of tech subjects for WIRED, in print and online, and has been contributing to the magazine since its inception. His weekly column, Plaintext, is exclusive to subscribers online but the newsletter version is open to all—sign up here. He has been writing about… Read more
Editor at Large

More from WIRED

Google DeepMind’s CEO Says Its Next Algorithm Will Eclipse ChatGPT

Demis Hassabis says the company is working on a system called Gemini that will draw on techniques that powered AlphaGo to a historic victory over a Go champion in 2016.

Will Knight

A New Attack Impacts Major AI Chatbots—and No One Knows How to Stop It

Researchers found a simple way to make ChatGPT, Bard, and other chatbots misbehave, proving that AI is hard to tame.

Will Knight

The Inventor Behind a Rush of AI Copyright Suits Is Trying to Show His Bot Is Sentient

Stephen Thaler’s series of high-profile copyright cases has made headlines worldwide. He’s done it to demonstrate his AI is capable of independent thought.

Will Bedingfield

The Myth of ‘Open Source’ AI

A new analysis shows that “open source” AI tools like Llama 2 are still controlled by big tech companies in a number of ways.

Will Knight

The World Isn’t Ready for the Next Decade of AI

Mustafa Suleyman, cofounder of DeepMind and Inflection AI, talks about how AI and other technologies will take over everything—and possibly threaten the very structure of the nation-state.

Gideon Lichfield

The AI-Powered, Totally Autonomous Future of War Is Here

Ships without crews. Self-directed drone swarms. How a US Navy task force is using off-the-shelf robotics and artificial intelligence to prepare for the next age of conflict.

Will Knight

How to Use AI to Talk to Whales—and Save Life on Earth

With ecosystems in crisis, engineers and scientists are teaming up to decipher what animals are saying. Their hope: By truly listening to nature, humans will decide to protect it.

Camille Bromley

In the Dreams Favela, Wi-Fi and Ecommerce Promise a Better Future

A tech-led initiative to bring economic development to Brazil’s shantytowns starts with a radical idea: Listen to the people who live there.

Angelica Mari

*****
Credit belongs to : www.wired.com

Check Also

Some People Actually Kind of Love Deepfakes

AI fakes are a disinformation menace. But some politicians, executives, and academics see them as …