Tom Nicholson

@elonmusk If human lifetimes are extended, maybe it doesn't matter. We have AI to come up with new ideas, so necrosis of thought doesn't matter that we don't die. This is neither good nor bad, it just is

@arankomatsuzaki Re parallel agents: tree of thoughts type branching etc? At every decision point, try them all and see what the outcome of that step is, rinse, repeat, and backtrack

i've been running Codex for ~8-24h per open math/physics research problem. few thoughts: parallel agents don't seem to scale that cleanly for a lot of problems. many of these are just extremely sequential. you don't really get to "spawn 50 agents and solve it from nowhere." it's more like: tiny move, check, reframe, tiny move, dead end, try again. hours/days of serial cognition, which honestly rhymes with how these fields move over decades. this updates me a bit against the sci-fi picture of "superhuman math/physics intelligence" as some alien oracle that instantly sees the proof / theory. the actual superhuman-ness is more mundane and maybe more important: the agent has absorbed a huge prior, can read long papers basically instantly, can think/write at >50 tok/s, and you can clone it across dozens of problems. speed + knowledge volume + multiplicability. that's the superpower. also: frontier physics seems much more tractable for these agents than decade-old open math problems. for some physics directions, ~8h is enough to get something paper-shaped and nontrivial. big caveat tho: research taste is still missing. the agent is a pretty good problem-solver, but not yet a top-tier problem-picker. it can push hard once the direction is chosen, but you probably still want a human with taste choosing the problem / framing / bet. current model: agents are becoming very strong research labor, but the bottleneck shifts upward into taste, problem selection, and knowing which hill is worth climbing.

@deanwball If there are no more algorithmic and data improvements (a fairly ridiculous assumption), the pace of change will be fast due to effects hardware advancements/ investments

I feel us approaching yet another summer of discontent with ai, just like last year, when many of my peers in the ai commentariat declared deep learning to have hit a wall because of gpt-5 blah blah blah.

@WillrichOstmann @pushmeet @GoogleDeepMind Now we just need to prove Lean is correct

@pushmeet @GoogleDeepMind The breakthrough isn't "AI solved 9 Erdős problems" — it's that the proofs are formally CHECKABLE in Lean/Coq. Not natural-language math you have to trust. That's why the 56-yr-old solutions are credible. Standard LLM math claims fail this bar.

AI agents are advancing research-level math. 🚀 I’m thrilled to share @GoogleDeepMind’s AlphaProof Nexus - an agentic framework for formal proof search powered by Gemini. When applied to a set of open formal math problems, our agent autonomously solved: ✅ 9 open Erdős problems (including two open for 56 years!) ✅ 44 Online Encyclopedia of Integer Sequences (OEIS) problems ✅ A 15-year-old open problem in algebraic geometry ✅ A 7-year-old open question in min-max optimization We are collaborating with mathematicians across disciplines - from combinatorics and graph theory to quantum optics. Ultimately, these results show the massive potential of even simple agentic loops powered by Gemini. Read the paper here: arxiv.org/abs/2605.22763…

@8teAPi People say that every year, we still have a

you think you’re down bad.. there are like 10 startups that raised money to build an AI mathematician and OpenAI just used a general model to go beyond the frontier… so what do you do now ? try to position for an acquihire ? because the clock is ticking… you probably have till the end of the year at best

@jxnlco Front-end prototyping

When do you reach for other models instead of Codex? What can we do better? Hit me with all of your frustrations. dms open. If you can give me detail (e.g. specifics/transcipts) - it'll help a lot in finding out exactly what we need to do to improve the next model

@thsottiaux Linux version

For those of you living inside the codex app, what should we prioritize among features, reliability or performance?

@kimmonismus iOS only no doubt

Looks like we got an answer to that cryptic openai post. codex mobile app. cant verify, hope its real :) would be really cool to see!

@sama Cost and speed

what would you most like to see improve in our next model?

@thsottiaux Show Linux some love!

As a Codex user, which platform are you on

@OpenAI Linux?

Codex now works directly in Chrome on macOS and Windows. It’s even better at working with apps and sites in Chrome, and now works in parallel across tabs in the background without taking over your browser. To get started, install the Chrome plugin in the Codex app.

@chrisgpt @basedjensen *yet

@basedjensen Remember @basedjensen 200-300k people a year get infected with Hanta in Asia and Europe But this is the first human to human multi person case Still not air born like Covid though

Aw shit ... Here we go

@thsottiaux Option to use pro. Grammar constrained decoding. More explicit search functionality

We are seeing strong traction and working to improve Codex for scientists across mathematics, physics, chemistry, biology and more. What do you wish it were capable of that it cannot do today?

@WorldsStrongest "Reverse parked my 18-wheeler first time, get in"

Caption this 📸

@TodayinHistory The most intelligent thing I've heard out of the White House for a long time

This may be the most articulate response I’ve ever heard to this question.

@aidan_mclau @pigeon__s Work on the personality

@pigeon__s what can we do to close the gap?

ok so gpt-5.5-instants personality seems to be like 3598x better than 5.3s and genuinely not slop i would say but i still feel like i prefer claudes a lot and even kimis but its like good enough now that i dont really care that much

@IterIntellectus Not irrational, just have bounded rationality, which is also modelled with game theory

the problem with game theory is that it assumes rational players but most humans are irrational

@SimplyKatie___ Learn from them

What’s the best way to handle someone who’s more knowledgeable or intelligent than you?

@oscpmentor @tenobrus He's not some dude, he's the dude

@tenobrus Did that guy ever turn out to be anything important? Is he actually an insider somewhere? Or just some dude?

so... where's gpt-6 strawberry boy??

@iruletheworldmo Rips through tokens like a laxative

codex feels genuienly remarkable with 5.5, it's hard to overstate just how much this level of intelligence with this level of speed and efficiency changes things. it's picking up nuances and understanding my intent far far better than any other model and rather feeling like im wrangling my way through one frustrating bottleneck after the next i'm coming away from sessions feeling genuinely delighted please just point this thing at something outrageous and see if it will do it, you may find yourself blown away by the results

@Hitchslap1 I'm doing it right now, so definitely

Serious question. Do you believe time travel is possible?