This week Mark Zuckerberg published 6,500 words on how AI is going to empower you, and the dominant reaction was a shrug attached to a question: from him?
Prose alone can’t bridge that gap for Meta. “The Future is for Everyone,” as the essay is called, pledges that “everyone will have an exceptionally capable personal agent that understands you, your goals, and everything you care about.” But the thing sitting in everyone’s memory is what happened to the last grand Zuckerberg promise.
The messenger problem
Kirsten Korosec, Rebecca Bellan and I dug into all of this on the newest episode of the Equity podcast. Our colleague Russell Brandom had already argued in print that the manifesto is “exactly why people don’t like AI.”
Bellan held Meta’s track record up next to the sales pitch. “You look back to the social media days, [Zuckerberg] was saying that he wants to make sure that everyone has an outlet for talking to their friends and having a social network,” she said. “And what do we have instead? We have ragebaiting and advertisements, and not connection.”
Two sentences, and the entire problem is laid bare. A vision document lands only when whoever is waving it has a history of actually delivering what it describes.
Meta isn’t in the frontier conversation, so it’s changing the conversation
To Bellan, the essay reads as a repositioning play. “Cynically, I think that this is an attempt for Meta to win in a different way,” she said. “They’re not winning in the frontier, closed-model space. They’re not necessarily even winning in the open space. But when it comes to personal empowerment, as Mark Zuckerberg talks about in his letter, ‘The Future Is for Everyone,’ that’s where he’s trying to win. He’s trying to provide the models that people will use for their own personal AI on their own personal devices.”
The previous edition of this letter, “Futures for Everyone,” aimed the AI squarely at glasses and wearables. The new one keeps its options open. “It’s not really clear what the final hardware form factor for this would be, and I guess he’s like leaving it up to the individual to choose,” Bellan said.
As for Meta’s new AI model, Glimmer, here’s the promise: “you’re using Glimmer for managing your schedule, drafting messages, organizing files, it’s always on, it can operate anytime, anywhere, with or without an internet connection,” she said. Muse Spark, meanwhile, remains tucked out of sight as the paid tier. “So that maintains a way for them to have some level of control over, still, its most capable models, and provide a revenue outlet for people who want to scale their compute, and people who want to like do larger projects, or companies that might want to do larger projects.”
I asked whether this is a land grab. The answer was yes
The way I put it on the show: Meta has poured serious money into AI, and a year ago there was a steady drip of stories about the sums it was dangling in front of AI scientists. Yet when anyone rattles off the frontier labs, Meta isn’t on the list. Consumer chatbots are the same story. Plenty of people do bump into Meta AI tools inside Instagram and Facebook, but Meta isn’t the name that springs to mind when you go shopping for a personal assistant.
Bellan agreed, though she flagged the recurring pattern. “Mark Zuckerberg loves to put out a vision of what his companies will do for people and how they will empower people,” she said. “Of course, Llama is great; it’s a great tool and open source, but I think that on a consumer level, how are people using Meta AI? Well, there’s like a lot of creepy-ass chatbots.”
Then she tried to actually download it
Of everything discussed, this is the bit that lodged in my head — the lone hands-on data point in the conversation, and it works directly against the essay’s own title.
“I was like, ‘Oh, maybe I’ll try to download Muse Glimmer and see what I can do with it,'” Bellan said. “Can I download it to my MacBook? No. You need specific hardware. I forget what the hardware is. It’s not actually that accessible to the average person.”
Korosec delivered the episode’s most compact verdict in reply. “It’s not for everyone. Not yet.”
Pollyannish, and priced accordingly
On why the backlash travelled so quickly, Korosec’s diagnosis was that “the manifesto didn’t strike the right note for a lot of people.” She circled back to Brandom’s column, calling it what’s “making people hate AI or hate on AI.”
“It is coming from Zuckerberg, and I do think also it feels very Pollyannish in a way,” she said. “He’s put these thoughts out there in the past, and it really doesn’t provide what I think is a more realistic picture, but instead casts AI as this great tool for humanity. That might be the case, but there will also be extreme costs to that, and I think that’s why the negative reaction to it has been so widespread.”
The anti-Dario positioning
In Bellan’s reading, Zuckerberg is claiming the corner directly opposite Anthropic CEO Dario Amodei, who spent the weekend rejecting the AI doomer label. Zuckerberg is “almost like an anti-Dario,” she said.
How she characterised the mood at the frontier labs: “‘We’re pacing the frontier, pulling back, focus on safety. We need to slow down development.’ They’re not doing any of that, of course, but they’re talking about it, and because of all these recent cybersecurity incidents that we’ve had.”
And Zuckerberg’s rebuttal, as she framed it: “No, no, we do not slow down. We cannot afford to give an inch to China, and slowing down would only hurt the individual, who can be empowered by this technology.”
One line in the manifesto I’ll defend
The essay is longer than nearly everything we’ve ever run. Even so, one sentence in it earns its place: “If you are an AI doomer and you believe that that’s the future that AI is going to bring, why are you building this stuff?”
A fair question. That doesn’t lead me to conclude the AI future is great, though. It leads me to raise an eyebrow at anyone who makes forecasts like these and then charges ahead regardless.
“Who wants this?”
Some of the pushback is about the messenger, and there’s no shortage of grounds for treating Zuckerberg and Meta with suspicion. Expect that chorus to swell in the coming months as the new Facebook movie hits screens.
The rest of it, though, is about the future on offer. That’s the engine behind the viral dogpiles that follow every time Sam Altman or Zuckerberg or anyone else sketches out daily life with AI, only to be met with a blunt “Who wants this?” Exhibit A: Altman musing that you could have ChatGPT generate podcasts about your kids’ interests to play on the school run. The universal response was, “Why don’t you just talk to your kid?”
One half of the manifesto is abstraction about unleashing creativity and invention, and abstraction wins over exactly zero skeptics, because it doesn’t land as real. The other half is concrete — the personal coaches and assistants — and parts of it are genuinely appealing. Plenty of it isn’t.
Want to stress-test the empowerment claim for yourself? Do what Bellan did. Pull up the Glimmer download page on the laptop that’s already in front of you and see how far you get.