Adrien Ecoffet

Some interesting ideas in there - glad people are looking at insurance markets, seems like an important bit neglected angle in AI governance. Stellar author list too! underwriting-agents.com

@StefanFSchubert Wonder if it will help Italy qualify

I think it’s probably right, I think there were fewer one-sided matches than feared. And it would get rid of the third place-qualification issue. Important that the new spots are distributed reasonably, though. Eg Concacaf shouldn’t get more.

(No criticism intended to the people who made these comparisons, thanks for contributing to the discussion!)

Today I learned that one Sam Altman is as persuasive as between 1 and 100 LBJs. Follow the thread for more! x.com/AdrienLE/statu…

Well I don't think Sam has mind control capabilities. What would need to be involved here are a lot of land acquisition issues, convincing legislators, convincing electors, these sorts of things, probably some environmental permitting things. It is a challenging task. Maybe a bunch of Sam clones could do it but that's not obvious to me. In some sense this is an appropriate comparison because arguably Elon has tried to cut the cruft in this project with the Hyperloop idea and that went nowhere. I realize it's a bit of a different take on it but it's not obvious that top entrepreneurs could achieve this. I think there is kind of a selection effect here at play where people don't get into politics because achieving these types of outcomes is so much harder than in the private sector

@AdrienLE I mean fairly close? Like I think that if ten Sam Altmans tried to get this done in under ten years and they had a 50B budget, and they had like v strong personal incentives to do this, I think they would succeed, and so I think a sufficiently persuasive AI would succeed

Could a superpersuasive AI get a high speed rail line between SF and LA constructed for under $50B and under 10 years with existing construction technology?

England seeing the rest of the world supporting them on Wednesday.

Would you say that it is because a sufficiently persuasive AI would have individual mind-control levels of persuasion (implying that this is a level of capability achievable by AI, which may be true but is a contentious issue), or because it could do this some other way (this other way may be contentious, at least some other people don’t see it it seems)? x.com/AdrienLE/statu…

@AdrienLE I mean, a sufficiently persuasive AI obviously could. I don’t understand how this is a contentious issue.

come for the best model, stay because we don’t treat you with contempt

Rest assured that GPT 5.6 Sol will stay in the ChatGPT subscription you pay for. Including Go, Plus and Pro subscriptions. At least until we ship an even better model.

@deanwball @__nmca__ To be clear on the prioritization point, my question was about "could" as in "would this be possible?" rather than "could" as in "could we please do this?"

@__nmca__ @AdrienLE obviously there are no guarantees, but “frontier ai systems reforming ceqa and then cleanly funneling tens of billions of dollars through California’s government” doesn’t seem realistic + like a thing to prioritize compared to many other things

[warning, shillpost] Had to read this twice because it so shockingly mischaracterizes (my experience of) OAI. I simply cannot overemphasize how wrong this is (again, from my experience). There have been multiple times, in both 1:1 and group settings, where I have personally directly disagreed with, corrected, or expressed frustration with leadership to Sam. Sam always responded with curiosity, open-mindedness, and even deference when I've brought disagreement/complaint/correction to him. Most times he has actively asked that I follow up with more thoughts or ideas on how we can do better on the subject. To the point of "street cred", at least one of these instances was in my first ~6 months at OAI, and nonetheless I was impressed by how quickly Sam changed his view when presented with data disagreeing with it (can clearly remember handful of people in that meeting who could attest). No place is perfect; OpenAI is obviously not a perfect company and Sam is not a perfect leader; he'd be first to admit that and has spoken in the past to places where he can improve. That seems healthy. We (leadership included) have made mistakes; but, as Steve Jobs would say, at least that means we made some decisions! To claim that OpenAI has a culture where people are retaliated against for honest criticism of leadership or overall company direction is (in my experience) truly ridiculous. We make fun of our chaos (yes, including leadership, which does have a very hard job) as much as anyone! I would say we do however have very little patience for talking a big game without delivering.

there will be people supporting england that have never supported england before

sam altman is probably the greatest ceo of our time. he practically has every major tech CEO investing hundreds of billions of dollars to compete with him, sometimes even colluding (see elon + zuck). I don't know if gates, zuck, or even elon had this level of competition when they were forming their companies. OpenAI still consistently churns out some of the best models and still emerges as the winner in almost every category. my only question is what did @paulg see during that 10 min YC interview

Kinda cool that tibo can reset the usage limits at ant now too

I have been thinking about the famous chart showing how experts keep projecting linear growth in solar installations, year after year, and always get it wrong when growth is still exponential. I think the same thing is happening with the discourse on product strategy around AI.

Updates for Codex and ChatGPT Work users. No nerfing, only good stuff! - We have landed inference optimizations and are passing down savings to all the subscriptions for GPT-5.6 Sol. That should result in around 10% more usage on its own. - We noticed that by changing the context size limit in the product to 372k for GPT-5.6 Sol, up from 272k for GPT-5.5, it resulted in more usage being charged than intended. We have reverted to 272k and will work to roll back out to 372k in the days to come. You should notice that usage drains significantly less after this change. - To understand where the extra usage was coming from, we ran some experiments where reasoning efforts were changed (referred to as juice values under the hood) and have reverted this. - There is slightly more usage of multi-agent than intended in high and xhigh reasoning effort, we are fixing this going forward. Also fixing a small other thing we noticed with auto-review where we can be more efficient. And we continue to have the 5h limit temporarily not apply. Enjoy the rest of the weekend!

@_____Lightning Also the source you sent claims it will take at least $30B, which is under $50B (and a very low estimate IIUC)

Not with that attitude ;). And I mean this kind of seriously. The project right now is looking to take around 3x the cost and time, but based on my admittedly cursory (ChatGPT thinking) research, it could cost around 50B if important tradeoffs were made such as not stopping at random Central Valley cities, limiting various forms of bureaucracy and environmental constraints, convincing owners who will need to give up land to do so and at fair prices, etc. Nothing completely crazy but a big need to focus the entire state on ruthlessly achieving the goal of connecting SF to LA rather than achieving a bunch of other constraints… basically a task for a superpersuasive AI.

@_____Lightning Tracks on the ground. And yes but it’d be convincing people to spend $50B on all this.

@AdrienLE Would it not be dependent on labor and equipment? Are we talking just laying tracks on the ground or tunneling ?

found an old screenshot of a conversation with Sydney