Meta has published “The Future is for Everyone: The Path to a Positive AI Future,” a 6,500 word ramble about what’s coming in AI — or what Meta would like to be coming. [Meta]

The piece is signed by CEO Mark Zuckerberg, but it it’s hard to tell if it’s chatbot output or just old-fashioned handmade artisanal corporate slop like we used to make.

Meta didn’t give us the prompt the blog post was generated from, so we have to decode what they meant. But you can be reassured:

Superintelligence will be among the most important technologies in history

We don’t have “superintelligence” and we don’t actually know what it is. But in this post, “superintelligence” seems to mean lots of chatbot.

The author-bot speaks against “extreme concentration of power.” This does not mean open weight models for everyone and computers with enough RAM to run the models at home. Instead, Meta proposes:

Everyone will have free or affordable access to these tools … We will offer free versions that will be accessible to billions of people.

That is, you live tethered to Meta’s chatbot.

The Zuckbot does want open models, though! Ones without those pesky copyright rules. Can’t compete with China with those:

Foreign labs currently hold several advantages here since American labs have to comply with many additional restrictions on training data.

Zuckerberg echoes Sam Altman’s prediction that after the singularity, you’ll still be working your crappy job. But you’ll have superintelligence to help you!

The Zuckbot insists Meta’s data centres are super nice now, and they use hardly any water and they build out so much cheap local power and they pay so much local tax. Because:

The main policy inputs into this are how efficiently we can build physical infrastructure

… Countries like China are bringing online 1GW+ of nuclear capacity every other week

They’re what? That’s AI.

Back here in reality, China connected 2.5 gigawatts of nuclear in the whole of 2025. There’s 36 nuclear construction projects in China, but the lead time is around a decade. China is not doing a gigawatt of nuclear every fortnight. The world is not doing a gigawatt of nuclear every fortnight. I guess the human took a break from the loop. [World Nuclear Report, PDF]

Meta’s real problem is that nowhere near enough people use its chatbot at all. So, like xAI, Meta wants to rent server space to Anthropic, in a $10 billion deal over two years — in the hope that someone will pay for all these data centres they built on bubblenomics. [NYT, archive]

This ridiculous essay is a plea for anyone to use Meta’s AI, which is mediocre even by the standards of AI bros. And not just run Llama at home at five tokens a second.

• Video — Podcast

Robot security guard have failed so hard that they’re being replaced with humans: [404, archive; Proof]

Proof News found evidence of at least 21 security robot deployments since 2015. We contacted contract holders and combed news articles and determined that at least 13 of those programs have ended.

Most of the cancelled programmes used robots from a company called Knightscope. You may remember the photos from 2017 of the security robot at a mall, the Knightscope K5 Autonomous Data Machine. It drove itself down some steps into a fountain, to the delight of Twitter. [NYT, archive]

In 2016, a Knightscope security robot at Stanford Mall in Palo Alto knocked down a toddler and ran over his foot. But it’s OK: [LA Times, archive]

The robots have had thousands of encounters with people and pets without incident, the company said.

Daxbot was publicly humiliated in January after Adam Pioth outwitted their campus security robot. He sat down, the robot came up and told him he was trespassing, and he said “hey, look over there!” and got up and spun around the robot, confusing it utterly. [Instagram, video]

Pioth spoke to Proof News:

the robots no longer shouted he was trespassing. Instead, he said, they would, “stop, look at me, and then just keep going really slow.” And then, he said, they were gone.

How is Knightscope dealing with its expensive robots not working?

The company is forging a new model, combining its robots and AI-powered software with another key ingredient: human security guards.

• Video — Podcast