Andrew Hart

I’m making amazing progress on building a new foundation for robot intelligence. I've skipped through a few placeholder names for my frontier research lab, but it deserves a serious name now. Introducing Base Intelligence. If you’re following me today, you’re in at the ground level, and I greatly appreciate that. You might remember my last startup, Hyper, where we solved the indoor GPS problem, and rolled it out with IKEA as the future of their shopping experience. Before that, I pioneered AR navigation (now integrated in Google Maps) and built the largest open source project for Apple’s AR platforms. When it comes to successfully developing new technologies and leading a lab, the number 1 requirement is a clear technical vision. You can’t just follow the status quo, or incrementally improve on what already exists. I’ve been saying that we need a new architecture for intelligence for almost 4 years. And now with Base Intelligence, I’m going to follow my vision and make it happen.

We contacted all xAI employees affected, but neither of them could be reached for comment.

The 2026 answer would be different, and unsurprising for Sam to learn.

Lunchtime walk in Belgravia, London

@kevinrose @digg It’s helped me disengage from X. High quality filter, I check in once a day and catch up. I love the focus.

a huge thank you to all my ai friends that have tried the new @digg over the last two weeks, we're off to a good start. more topics are on the roadmap 🙏

People pay more for a coffee from Starbucks, where the barista makes it themselves, than from Pret, where they push a button on a machine. Starbucks have the technology to automate, it’s the human touch that makes it more premium. Extrapolating that out, I predict that with improved robotics, many fast food & drink places will automate, but front-of-house humans will be a premium in some. In other words, you can get your Subway made by a robot for cheap, but the more expensive deli will always offer a sandwich made a human.

After reading this book about intelligence, written in 2004, I found this “mid” review in 2012 from one Andrej Kaparthy. Guess he was skeptical.

Written in 2004, decades ahead of the AI boom, and yet many of the ideas still remain under-explored. Like, what if we structure AI systems similar to the brain? Good dive into neuroscience. 4.5/5

@aaronp613 @ParkerOrtolani it's not a butthole thats for sure

@ParkerOrtolani Wtf is that third one

choose your team

THE LITTLE GUY WINS

@BenjaminCrew1 they often use Royal Naval College or Queens House in Greenwich, London, for lots of sets including the white house. Here’s some show they were filming in January (to look like an outdoor white house event at xmas)

@gail35zx @dimaip @darrenpjones ok

@dimaip @AndrewHart @darrenpjones It’s like when some argue, if you pay more ‘they will work harder’! Someone, somewhere make it make sense. e.g. doesn’t explain why so many third raters assume huge responsibility + how can you work harder just for higher rate/perks - impossible!

I’m looking for a new personal assistant. Based in the House of Commons, you’ll help me with my diary, including supporting my evening meetings on Mondays and Tuesdays. This role sits at the centre of my parliamentary, constituency, government and special advisor teams helping me to be as effective as possible. Deadline: 29th May. More details 👇 w4mpjobs.org/JobDetails.asp…

Excited to return to Amsterdam next week. It’s a working trip while i focus on my intelligence research, working from coffee shops (actual ones…). And would love to meet up with a few people while I’m there.

Watching a massive CGI battle in a big budget movie: 😐 Watching the scene in Willy Wonka and the Chocolate Factory where Nixon has Wonka bars delivered to the White House: HOW DID THEY DO THAT???

@ChristianSelig @tvn_dev @thevraa I’m unfollowing this thread and never want to hear the words “Infuse” or “transcoding” ever again

@tvn_dev @AndrewHart @thevraa No, even with Plex there's no support for direct play with Infuse unfortunately. A lot of AV1 files can play back fine, but get a beefy enough one and it hitches noticeably. H265 and 264 is easy though as Apple TV has hardware decoders for those

I'm convinced Jellyfin would pass Plex if there just existed a good Apple TV app

I changed the model down to 4.6

Claude Code is extremely slow today. Often taking several minutes to respond to each query with no visible activity. So I asked Claude to review the logs *ahem* apparently its just stalling for several minutes for no reason 😟

@Simon_Hypixel @LeadingReport it's time for someone to build a 2.0

@LeadingReport One of the greatest game of all time, changed the lives of many

Exactly 17 years ago today, Minecraft, the best-selling video game of all time, was released.

sounds a bit ambitious but good luck to him

@ChristianSelig @jroav @thevraa have you tried using it and decided it’s no good? what was your experience?

@jroav @AndrewHart @thevraa No, Infuse only direct plays, no transcoding occurs *anywhere*. It sends the file to the client and the client plays it however it sees fit, in this case on Apple TV a super slow software decoder. What I want *is* a transcoding option on the server, which Infuse ignores