Common ground between the authors of AI 2027 and AI as Normal Technology! Coauthored article below.

I have a lot of sympathy for the 'AI as normal technology' view. On the face of it, recursive self-improvement => superintelligence feels like an overspecific scenario leading to a sci-fi outcome. Accompanied by draconian governance interventions premised on an imaginary problem.

I interviewed Holden Karnofsky of Anthropic (and past CEO of Open Philanthropy) for 4.5 hours on almost all his AGI takes: • The AI 'race' isn't a coordination failure [00:18:01] • Why he's currently focused on pursuing easy wins and partial victories [02:43:43] • "It'll be

E/acc's Beff Jezos thinks we should build AGI faster because life accelerates entropy production, and that's supposedly good. 🧵 This is genuinely unprecedented in moral philosophy. Nobody – not Nietzsche, not the transhumanists, not even cosmic pessimists – ever argued that

Public opinion polling on AI in US+UK+France+Germany+Poland. Kill switches and liability for harms were the most popular policies — above evals even.

NEW on Transformer: The UK government today announced a £15 million "Alignment Project", which will fund research into AI alignment, control, and interpretability. The government called AI alignment "one of the most urgent technical challenges of our time".

Inference Scaling and the Log-x Chart: 2024 saw a switch in focus from scaling up the compute used to train frontier AI models to scaling up the compute used to run them. How well is this inference scaling going? 1/

The UK Government has nailed it again on the software side. This project is excellent: https://t.co/ApypXz4Dvm @AISafetyInst - thank you for your work!

New piece out with my colleague Zach Arnold, laying out some basic common-sense policies that people with a wide range of views on AI could get behind: -Building govt capacity -Better measurement -Independent auditing -Disclosure -Incident tracking -Liability 🧵:

A comprehensive database of AI risks 😍

This might be the best thing I've watched this year. https://t.co/UNQSuB1gdo

The Obsidian Canvas file format is now called JSON Canvas and has its own site, specification, and open source resources. JSON Canvas can be implemented freely as an import, export, and storage format for any app or tool. All the resources associated with JSON Canvas are open

Rather than allow other browser engines to install-to-homescreen on iOS, Apple is removing the feature entirely in the EU. Some features are tied to install-to-homescreen, such as push messages. Apple is unshipping those for EU users. We need to act now:

Apple banned some Beeper user Mac computers, New York Times investigated, Apple unbanned. The saga continues! Even after we've shifted our energy from iMessage back to our core Beeper product (15 network universal chat app), Apple cannot seem to get enough! Here's the TLDR:

What if instead of infotainment, news coverage was like the weather forecast? A dashboard of probabilities... Introducing v1 of the Trump dashboard, just in time for the Iowa caucus (v2 will have pretty charts)

I'm a little annoyed at React rn https://t.co/z79OJYd5cj

Average cost of a medium sized car is $13,000 per year. What else could that get? I did some maths:

I'm hiring research engineers! If you're an engineer interested in building good, honest AI models, apply and mention honesty as an area of interest :)

Global Shield is #hiring! Come join us in an "all of the above, jack-of-all-trades" role that allows for tremendous personal growth for the right candidate. Join us now in the early stages of our journey to reduce global catastrophic risk, globally! https://t.co/ASwpAnkqpV

lots of predictable takes and hypotheses that magically align perfectly with posters' preconceptions and narratives about safety or accel or open source or whatever they care about. in the absence of any information maybe just wait