Zuckerberg's AI-for-all pitch faces a big problem

Mark Zuckerberg says AI is great for everyone — but just to make sure you believe him, he's going to spend $1 billion in communities where he's going to build AI data centers. AP Photo/Nic Coury The guy who got everyone using social media now wants everyone on board with AI.Mark Zuckerberg made the case for AI-for-all in an impassioned, 6,500-word blog. (Exhibit A: People who need editors.)Zuckerberg's post laid out his vision for an AI future and why he doesn't think we should all be afraid of it.He joins a growing cohort of tech executives, including Anthropic's Dario Amodei and Microsoft's Satya Nadella, who are writing lengthy essays on AI's impact on the future.Zuckerberg's post wasn't random. It coincided with Meta's launch of a new agentic AI model, Muse Glimmer. The model is open-weight, meaning users can download and modify it for free on their own computers. More details on all things Glimmer.His AI opus highlights the growing debate in the tech community over whether AI models should be open or closed.Supporters of open-weight models say allowing developers to tinker benefits the entire industry. Those in favor of closed models say the tech is too powerful to make it widely accessible. Both arguments have their flaws. If your company doesn't have one of the top models, it's easy to support opening up models for everyone. You're not giving up much of a competitive edge.Meanwhile, top AI companies are struggling to contain their top models. That raises the question: Can you argue you should be the only ones in charge of your models when you're not doing a good job of controlling them in the first place?Zuckerberg's AI-for-all pitch faces a big problem.People have strong feelings about AI and the infrastructure that supports it, such as data centers. One way he's trying to get around that is by giving out lots of money.Meta is creating a $1 billion fund for communities where it is building AI data centers. (Hat tip to my colleague Peter Kafka, who suggested Meta do something like this a few weeks ago.)A 6,500-word blog post. Multiple media stops. A $1 billion fund. That's a lot of time and effort to get people using AI agents.But what other choice does Meta have?The company is built on everyday people using its products (Facebook, Instagram, WhatsApp) and then selling companies access to their attention. It's not as if Meta can completely upend its business model and start selling AI automation tools to companies.In some ways, that's fine. AI agents can be a good fit for Meta's user base. The company has a ton of data on individuals that AI can use to become their highly personalized assistants.But you've got to get people comfortable with using AI in the first place. Dan is the lead writer for BI Today, Business Insider's flagship daily newsletter. Dan often interviews executives about everything from AI's impact on capitalism to robotics to the potential SaaSpocalypse as part of his work on the newsletter.Dan was an editor and reporter at BI, covering financial technology and market structure.His previous work includes everything from inside Robinhood's failed "Checking and Savings" product that eventually led to Congress getting involved to the internal arguments over JPMorgan's failed attempt to launch a finance app for millennials.Before joining Business Insider, Dan wrote about risk management in derivatives markets for Risk.net and fintech for WatersTechnology. He initially covered local sports for The Journal News, a daily newspaper serving the lower Hudson Valley. Got a tip? Contact this editor via email at ddefrancesco@businessinsider.com. Newsletters Newsletter