Thomas Larsen (@thlarsen) / X

Thomas Larsen

139 posts

Thomas Larsen

@thlarsen

Researcher at AI Futures Project. AGI is going to be a really big deal, we don't know when it's going to happen, and we're not ready for it.

Joined August 2022

Thomas Larsen
@thlarsen
Jun 19, 2025
Lots of people in AI, and especially AI policy, seem to think that aligning superintelligence is the most important issue of our time, and that failure could easily lead to extinction -- like what happened in AI 2027. But they don’t mention this fact in public because it sounds
40K
Thomas Larsen
@thlarsen
Apr 3, 2025
I think we're on the cusp of AGI, and society has barely begun to think through the implications. We wrote an extremely detailed scenario to explore what the AI transition will look like.
Daniel Kokotajlo
@DKokotajlo
Apr 3, 2025
"How, exactly, could AI take over by 2027?" Introducing AI 2027: a deeply-researched scenario forecast I wrote alongside @slatestarcodex, @eli_lifland, and @thlarsen
9.2K
Thomas Larsen
@thlarsen
Jun 16, 2025
AI 2027 skeptics can't give a coherent scenario for how they think AI progress will go. I would love it if @GaryMarcus, AI is a normal technology folks, @MechanizeWork people, etc, would actually write out what they think will happen
20K
Thomas Larsen
@thlarsen
Jun 27, 2025
Want to get up to speed on AI? My top recommendations are: - AI 2027 - Situational Awareness - AGI Ruin: A List of Lethalities - OpenAI Email Archives (from Musk v Altman) - Binging all the AI-related Dwarkesh podcast episodes.
7K
Thomas Larsen
@thlarsen
Jun 5, 2025
I think AGI/ASI skeptics should be much more specific about what level of AI capabilities they are confident we won't get in the next 5/10 years? For example, in the AI 2027 takeoff post -- which milestone will we be blocked by?
3.2K
Thomas Larsen
@thlarsen
Jun 20, 2025
This claim from "AI as a normal technology" is clearly wrong, and I'm disappointed that it has gotten so much traction. 1. A lower bound for the capabilities of ASI is like a human, but sped up by a factor of 100x and working 24/7. 2. This would already clearly be
4K
Thomas Larsen
@thlarsen
Jul 8, 2025
The main sycophancy threat model is that humans are imperfect raters, and so training AIs with human feedback will naturally lead to the AIs learning to produce outputs that look good to the human raters, but are not actually good. This is pretty clear in the AI safety
Arvind Narayanan
@random_walker
Jul 7, 2025
A few people have asked me if a technical fix for AI model sycophancy is on the cards. In fact, a technical fix for sycophancy is trivial. In many cases all it would take is a tweak to the system prompt. The reason companies are struggling to get this right is not technical.
14K
Thomas Larsen
@thlarsen
Jul 30, 2025
We're right on track for the AI 2027 revenue prediction. $12B/yr annualized revenue for OpenBrain projected for Aug 2025
morgan —
@morqon
Jul 30, 2025
anthropic is suggesting $9 billion of revenue (annualised run-rate) by the end of this year, more than double their previous “optimistic” forecast of $4 billion thank you, claude code
3.1K
Thomas Larsen
@thlarsen
Jun 28, 2025
I agree with Eli that these are important areas. But IMO the most important jobs in the world probably aren't on this list, instead, they are things like: - Starting a new org to fill a huge whole in the AI safety ecosystem. - Getting a job that could impact the overall USG
Eli Lifland
@eli_lifland
Jun 28, 2025
Replying to @eli_lifland
There are many ways to use one's career to help AGI go better, here we list some of the top ones.
3.9K
Thomas Larsen
@thlarsen
Jul 8, 2025
I like thinking for myself, so I try to never defer to anyone. But if I did, I'd defer to Ryan. Worth listening to, many important considerations discussed here.
Rob Wiblin
@robertwiblin
Jul 8, 2025
Ryan Greenblatt is lead author of "Alignment faking in LLMs" and one of AI's most productive researchers. He puts a 25% probability on automating AI research by 2029. We discuss: • Concrete evidence for and against AGI coming soon • The 4 easiest ways for AI to take over •
00:00
7.1K
Thomas Larsen
@thlarsen
Jun 25, 2025
Great post on one of the AI 2027 TTXs! I strongly agree with "The biggest threat to a rogue AI is … other AI?".
Steven Adler
@sjgadler
Jun 24, 2025
New post! A crisis simulation changed how I think about AI risk
5.6K
Thomas Larsen
@thlarsen
Jun 6, 2025
I think it's underrated that AIs will have to solve their own alignment problem. There's a good chance this will be difficult for even slightly superhuman AIs, and I think this increases the general amount of chaos around the time of AGI (e.g. egregious warning shots).
1.9K
Thomas Larsen
@thlarsen
Aug 7, 2025
Replying to @romeovdean
Looks like they had the correct graph on the website, so they must have intentionally decreased the o3 bar for the livestream???
1.7K
Thomas Larsen
@thlarsen
Aug 7, 2025
surreal how bad this graph is
Romeo Dean
@romeovdean
Aug 7, 2025
52.8 > 69.1 = 30.8
1.1K