
building pab, the AI co-researcher that does real research. every researcher, the manpower of a entire lab @propabAI
Joined July 2026
- Tweets671
- Following135
- Followers137
- Likes3.8K
Pinned Tweet
every field is moving faster than any researcher can keep up with.
and every researcher i know has 100s of ideas but only time to test 2-5, the rest just die in a notes app.
propab is my answer to those: an AI co-researcher that takes your idea and actually runs with it.
- searches ~300M papers
- forms hypotheses, writes and runs real experiments
- connect to your compute/cluster
- shows the evidence behind every claim
early beta opens today : propabai.com
afaict the shortest path to faster science is letting researchers test all 100 ideas, not 2.
Shani Singh retweeted
1. LLMs can't even do multiplications.
2. Ok they can do multiplications but they can't do math competitions.
3. Ok they can do math competitions but they can't get gold at them.
4. Ok they can get gold but it's not like they can solve known unsolved problems.
5. Ok they can solve some known unsolved problems but not millennium prize problems.
6. Ok they can solve a millennium prize problem but it's not the Riemann Hypothesis.
7. Ok they can solve the Riemann Hypothesis but...
2027 AI x chemistry discoveries prediction:
a) room temperature ammonia synthesis at atmospheric pressure
b) catalysts converting co2 directly into methanol
c) sodium batteries exceeding lithium energy density
d) enzymes breaking down forever chemicals in hours
e) catalysts splitting seawater without toxic byproducts
f) antibiotics targeting previously untreatable bacterial mechanisms
g) carbon capture materials regenerating without external heat
h) catalysts converting methane directly into liquid fuels
i) biodegradable polymers matching conventional plastic strength
my guess is 3-5 of these by end of next year:)
2
4
105
recommend read out of all the hype.
We're sharing new data on how AI is changing research at OpenAI.
We find researchers' AI use continues to rise exponentially, and agents are completing longer tasks, but human involvement remains central in research. Researchers are also spending more time on safety and alignment, with human effort in these areas growing faster than in any other individual research area.
This report aims to increase transparency around research acceleration and inform the governance of increasingly capable AI. We believe that AI labs should be required to track and publicly report indicators of recursive self-improvement (RSI): the process by which AI accelerates the development of future AI systems.
Our hope is that this expanded report will help sustain and advance shared measurement norms across the industry.
openai.com/index/research-ac…
1
29
seems like google has a model production pipeline now that is shipping model every week
are these just newer RL checkpoints as different models or what? cz pertaining entire model isn't possible every week nor feasible
Gemini 4 Neon is rolling out now.
Sitting between Carbon and Hydrogen, it brings frontier intelligence and fast task completion for complex software engineering, knowledge work, and science.
2
125
its frustrating to use fable 5.2 /opus 5.5 thesedays due to the classifier that keep rejecting normal chats too
intelligence getting controlled by AI GOD's finally
2
207
eventually people will rent out their entire digital identity to ai agents and become the biological backup account
1
1
106
let's be honest, super intelligence or not but even this version of AI is very useful
not to compare that entire gen AI is just a by product of humans trying to make ml do machine translation better
Everytime I see a Dwarkesh video, I'm wondering what they are smoking in SF. It seems very distant from reality
"Superintelligence" is one of the cringiest terms I've seen, it's a rebrand of AI to get more money out of VCs.
An no, recursive self-improvement hasn't started. It's just human researchers prompting to say "speed it up". It does some cute kernel optimizations and hyperparameter sweeps, very similar to the NanoGPT speed run records done by humans. Nothing fundamental or groundbreaking gets invented by so-called superintelligence.
To give an example: when @Recursive_SI released their (impressive) Flint model on @huggingface, I cringed a little for their use of the terms "superintelligence" + "recursive self-improvement". As said in their blog post, through "recursive SI", LLMs were able to figure out - I kid you not - to determine how many total parameters and active parameters the model should have. So yeah, it's just changing a config. It's doing hyperparameter sweeps on things humans have built (like MoE's, validation loss, perplexity etc). Of course, you shouldn't tell that to VCs. You should scare them by saying AI is improving itself at an alarming rate - this increases the $$$ raised by another factor.
Can we please stop all this misinformation: no, superintelligence is not here. No, recursive self-improvement hasn't started. And yes, there's lots of work to do to build more reliable AI that is actually smart.
This is why there's such a huge gap between super impressive benchmark numbers of LLMs, and the (lacking) economical impact of them. It's not able to do research like a human, doing experiments, coming up with actual new stuff, and generalize beyond their large amounts of training data.
2
246
need sponsorship for 8x H200/B200 equivalent GPUs for ~2 weeks for doing some cool research experiments
happy to oss later, dm.
one day the biggest competitive advantage in science might be owning 10000 autonomous laboratories rather than employing 10000 scientists
1
69
I certainly believe we might end up creating ai scientists that discover things humanity cannot understand and then spend decades building instruments just to verify their predictions
3
71
it feels like the simulation has all of sudden been accelerated?
most uncomfortable and equally exciting moments are when I traceback the development of AI and somehow I'm internalised the overall acceleration like it's not a big deal
deepSeek r1 just released in Jan 25 and models were struggling in basic high school mathematics and multi file debugging
then how the hell we landed up with 10k parallel agents solving millenium problems and 1k agents writing entire operating system in just 21 months???
i mean literally whole world is going to change in entirely unexpected way, governments will collapse, brutal layoffs are incoming across every major IT sector, economics and labors across countries might be drastically affected, and so much more I'm not able to comprehend
and yet somehow people outside X are unaware of this massive hit that will affect their entire day to day work and identity
and even governments seems ignorant on this, obv I'm not a dommer and I love AI but governments should start critical critical awareness program to help people know about the upcoming events.
all the believe and assumptions humans had held for centuries about medicine, illnesses, possibly even death would be stale in upcoming years with cure to all diseases found on same day/week not 20 years of research
its easy to read and yet lookpast this but if you truely feel this then you'll be shocked as hell
this tech if I compare is more comfortable to humans learning about fire rather than electricity
why I feel like this? cz RL (method used for post training model) has technically no upper bound, secret sauce to this problems are continual learning as I belive, cz once model aggregates experiences then emegence of taste is a by product of it and hence thinks like reflection and counterfactuals will emerge naturally, so it's just the unlock of these models stateful systems instead of stateless
I truly believe superhuman coder models by end of 2027 is a likely outcome given the rate the progress isn't steered.
few of takes are my personal feelings and thoughts not a prophecy of that happening in that way.
1
116
Shani Singh retweeted
We've given all Pro, Max, Ultra, and Team users one free full reset, across all models and limits. Use it anytime before December 10 at 2:00PM PT.
We recently fixed a few small issues that used more limits than intended, and the Fable resets didn't do much for those of you who'd already hit your weekly limits for all models. While these issues didn't affect everyone, we want to make up for the trouble.
You'll find it in Settings > Usage, or at clau.de/reset.








