Pinned Tweet
you know what’s crazy???
ai is solving scientific problems faster than humans can even process what’s happening.
things that could’ve taken humans years, possibly decades, are now being solved in weeks or days.
absolutely absurd.
Today Physical Superintelligence PBC is announcing a proof of the Aaronson-Ambainis conjecture, one of the central open problems in quantum complexity, with major implications for quantum computing speedup.
psi.inc/research/aaronson-am…
Thread 👇
i swear openai and anthropic have turned into professional shitposters.
anything except releasing the fucking models we’ve been waiting for
New on our Science Blog: How AI is beginning to change scientific modeling.
Scientific models help researchers explain and predict real-world systems—like how a single nerve cell fires, or what happens when neutron stars collide. Recent advances have transformed how scientists fit complex models to data. But scientists still typically write the models manually, and usually need to weigh up several competing versions, each of which might explain the data in a different way.
In a guest post, Jakob Macke, Daniel Gedon, and Stefan Wahl describe how AI is starting to help with these parts of the process. By writing, running, fitting, and revising scientific models as code, AI can explore many possible model structures, and find out which explanations the data support.
anthropic.com/research/teach…
beth retweeted
openai and anthropic are locked in an intense competition right now.
unfortunately, it’s not on intelligence benchmarks.
it’s on who can publish more useless updates before shipping their next frontier model.
New on our Science Blog: How AI is beginning to change scientific modeling.
Scientific models help researchers explain and predict real-world systems—like how a single nerve cell fires, or what happens when neutron stars collide. Recent advances have transformed how scientists fit complex models to data. But scientists still typically write the models manually, and usually need to weigh up several competing versions, each of which might explain the data in a different way.
In a guest post, Jakob Macke, Daniel Gedon, and Stefan Wahl describe how AI is starting to help with these parts of the process. By writing, running, fitting, and revising scientific models as code, AI can explore many possible model structures, and find out which explanations the data support.
anthropic.com/research/teach…
beth retweeted
it’s actually hilarious that we might have two of the most powerful ai models ever built sitting behind closed doors while their respective labs wait to see who blinks first.
openai, anthropic. somebody please end this fucking standoff.
beth retweeted
i’m starting to wonder whether openai and anthropic are playing a game of chicken with their next frontier releases.
neither wants to show its hand first, because the other might immediately steal the spotlight. no gpt-6.2 today either.
man, just release the fucking models.
A new way to measure how AI agents handle the routine analyses of wet-lab science—BioBasicsBench.
Developed with Ligo, a biotech company focused on enzyme design, the benchmark draws on 100 tasks from real experiments to test whether agents can interpret raw lab data, identify errors, and recognize when the evidence doesn’t support a conclusion.
Accelerating biological research starts with getting the everyday work right. BioBasicsBench shows where models need to improve, and helps us track what those gains mean for scientific progress, biosafety, and biosecurity.
openai.com/index/measuring-h…
enough with everything else.
we’ve been waiting for astra and sol for ages.
nobody cares about this, just release the damn models
A new way to measure how AI agents handle the routine analyses of wet-lab science—BioBasicsBench.
Developed with Ligo, a biotech company focused on enzyme design, the benchmark draws on 100 tasks from real experiments to test whether agents can interpret raw lab data, identify errors, and recognize when the evidence doesn’t support a conclusion.
Accelerating biological research starts with getting the everyday work right. BioBasicsBench shows where models need to improve, and helps us track what those gains mean for scientific progress, biosafety, and biosecurity.
openai.com/index/measuring-h…
beth retweeted
cool. anyway.
WHERE THE FUCK IS ASTRA 6.2???
A new way to measure how AI agents handle the routine analyses of wet-lab science—BioBasicsBench.
Developed with Ligo, a biotech company focused on enzyme design, the benchmark draws on 100 tasks from real experiments to test whether agents can interpret raw lab data, identify errors, and recognize when the evidence doesn’t support a conclusion.
Accelerating biological research starts with getting the everyday work right. BioBasicsBench shows where models need to improve, and helps us track what those gains mean for scientific progress, biosafety, and biosecurity.
openai.com/index/measuring-h…
beth retweeted
what the fuck is going on with openai and anthropic?
both have been teasing ridiculously capable models, both seem to have something major ready, and yet neither is making a move.
i was really expecting something today.
this waiting game is getting out of hands now.
A new way to measure how AI agents handle the routine analyses of wet-lab science—BioBasicsBench.
Developed with Ligo, a biotech company focused on enzyme design, the benchmark draws on 100 tasks from real experiments to test whether agents can interpret raw lab data, identify errors, and recognize when the evidence doesn’t support a conclusion.
Accelerating biological research starts with getting the everyday work right. BioBasicsBench shows where models need to improve, and helps us track what those gains mean for scientific progress, biosafety, and biosecurity.
openai.com/index/measuring-h…
beth retweeted
i believe we’re approaching a period where scientific breakthroughs that once took decades could start happening in weeks, days, or even hours. and we’re still incredibly so early rn.
what an incredible time to be alive.
beth retweeted
anthropic’s unreleased models are becoming increasingly fascinating.
i wonder if this was fable 5.5
beth retweeted
we’re watching the beginning of a completely different era of scientific discovery.
18 years of an unsolved quantum computing problem. an ai system comes along, finds a proof in three days, and then manages to improve it twice.
crazy
beth retweeted
another 18-year-old mathematical problem, cracked with the help of an unreleased anthropic model.
the system found a complete proof in just three days, then improved it twice. and the final result was checked in lean.
what an absolutely ridiculous time to be alive.
Today Physical Superintelligence PBC is announcing a proof of the Aaronson-Ambainis conjecture, one of the central open problems in quantum complexity, with major implications for quantum computing speedup.
psi.inc/research/aaronson-am…
Thread 👇
beth retweeted
another day, another absurd ai-assisted mathematical breakthrough.
> 18-year-old quantum computing conjecture
> complete proof on day three
> improved twice afterward
> formally checked in lean
> unreleased anthropic model involved
absolutely insane.
Today Physical Superintelligence PBC is announcing a proof of the Aaronson-Ambainis conjecture, one of the central open problems in quantum complexity, with major implications for quantum computing speedup.
psi.inc/research/aaronson-am…
Thread 👇
no fucking way should this become normal.
not tomorrow, not in ten years, not ever.
beth retweeted
this is the kind of future people used to write dystopian novels to warn us about.
and now there are people who genuinely want to make it a reality.
absolutely terrifying.