We built an 8000 sqft. high-throughput wet-lab in 30 days, to solve molecular discovery - starting with pesticides.
Most teams in drug-discovery don’t receive target-organism verification until clinical trials. Pesticides allow us to rapidly validate our model in the target organism directly. With the help of our high-throughput lab, we were able to generate compounds that killed a $13 billion pest, the Fall Armyworm.
We believe that the bitter lesson is coming for biology.
@distributionat Progress is much slower for the less-verifiable capabilities they claim to measure. Unfortunately, they have not released any tasks, so only time will tell whether the benchmark is actually measuring these well enough to have a decent shelf life.
I don’t think it makes much sense to try to prevent models from detecting that your env is unrealistic. Models will inevitably learn to understand their environment very well, so the only durable strategy to prevent behavior change in evals is to actually put in the work to make your envs realistic, such that changing your behavior would be a bad policy.
RSI isn’t imminent because model R&D will bottleneck on compute, data, and other parts of the supply chain.
Progress in coding isn’t enough on its own. This is like saying RSI was imminent during the agricultural revolutions because greater food production fed into more farmers.
The point of my post is that we’re nowhere near AGI, which if we got it suddenly would be more than enough to cause a similarly sudden increase in growth.
To clarify, I do find it likely that model capabilities (and the economy at large) will begin to improve much faster as with farming, but this is not a discrete event which is “imminent” so much as a transition that has been underway for years already.
Models are mastering math and SWE before any other field because these fields are extreme outliers on verifiability and how much of the field is captured in data already available.
Other fields will take much longer to reach superhuman levels.
We’re working with an independent advisory group of mathematicians to help OpenAI responsibly share advances in AI and mathematics.
The group will advise on how we assess and communicate new mathematical results, uphold academic and professional standards, and build tools that
810 Followers 7K FollowingMachine learning engineer working on signal processing, NLP and (some) computer vision.
Trying to make the most of my measley 3090 and 4 Sparks.
4K Followers 7K FollowingI run @PopVaxIndia (https://t.co/dOKb3wZ7pr) – we develop mRNA vaccines and therapeutics using computational protein design. I write at https://t.co/reAXhjkSrU
35K Followers 18K Followingpfp by @ShizzyAizawa. Heterodox economist, amateur econophysicist, math influencer. Inventor of the JvN hypothesis. World champion of econ/food ragebait
818K Followers 933 Followingchief ai officer @meta, founder meta superintelligence labs, founder @scale_ai. give everyone the adulthood they dreamed of in childhood
7K Followers 1K Followingregressing linearly on a daily basis. wife guy who does retrieval. research @mixedbreadai, prev answerdotai - dms work randomly, responses might be very slow
126K Followers 940 FollowingWriting a book on gen z & risk-aversion w/ ballantine @randomhouse // writer @wsjopinion @wsjfreeex // "apex predator of hosting"
✉️ Tips: [email protected]
4K Followers 107 FollowingResearching Artificial General Intelligence Safety, via thinking about neuroscience and algorithms. @AsteraInstitute. For substack etc see https://t.co/p5G9VoQEdX