Paul Wollenhaupt

Biology must be programmable. We will make it so.

Capped prophet

If @realDonaldTrump & Xi can pull off a deal that keeps humans in charge of machines for the long haul, it deserves a Nobel Peace Price. 🏆

@CFGeek Could you elaborate why?

Claim: for interpretability and monitorability purposes, this kind of parallel text generation is no worse than sequential generation

@47fucb4r8c69323 Nah, even when SBF had money, he didn't fund MIRI. I retrospect this should have been everyone's first clue that he was secretly a villain. What kind of person does that? What kind of human being?

Even though it causes me a lot of trouble, it’s quite a good thing that people care so much about being called immoral. The soul wants to be righteous. That’s cause for hope.

See you at ICLR! 🇧🇷 Keen to chat about diffusion models, pre-training, molecule & protein design and anything standout!

I'll be at ICLR in Rio this week. DM me if you want to meet up and chat AI x Bio!

@mgdurrant Curios, what kind of problems is this tracking? What are typical inputs and outputs, i.e. questions, actual structures, does it do conputational work on the structures or answer quesions with context?

Feel free to throw your hardest structural biology problems at Opus 4.7, and please share feedback so we can keep improving!

@SHL0MS Every life is valuable, including Sam's. I'm glad he and his family are okay. Attempting arson on a family home sounds evil because it is evil. Goodness is not a deontological luxury that melts away in the face of armageddon, it's vital to navigating challenges such as AI.

Introducing self-correcting search: a technique to let diffusion models self-correct mid-trajectory. Working with @RadicalAI, we gave MatterGen a feedback loop from its own activations, improving viable on-target candidates by ~30%. (1/8)

if we can't get our shit together i'm gonna have to stop "well actually"ing my friends who blame capitalism for ruining the world. the critique would actually fit for once. do you have any idea how embarassing this is for me. please. i'm begging you

There is a sort of fatalism in "someone will be hired to replace you" that limits any perceptions of agency over any market (e.g. if you don't buy the product, someone will) but markets are an aggregate, and every individual's decision in that aggregate does move it a little.

@danny_nkjg @Sauers_ Decorators do not need to return functions/classes @lambda x: x()

@Sauers_ Well this just shattered my day

"self" is just convention in Python, not part of the language. lol

things we need for a good AGI future: 1) the technical ability to align a superintelligence 2) sufficient coordination among most of the leading AGI efforts 3) an effective global regulatory framework including democratic governance

I think people deserve a good explanation as to why proper diplomatic measures haven't been properly tried if we're going to blatantly diagnose the issue with this disturbingly literal meme. It's a bit of a cuck move to simply let the techno capital machine eat your free will. I am disturbed by everyone's resigned acquiescence.

Dario said the same thing during The Day After AGI discussion this morning. They were both asked for their timelines: Demis said five years; Dario said two. Later in the discussion, Dario said that if he had the option to slow things down, he would, because it would give us more time to absorb all the changes. He said that if Anthropic and DeepMind were the only two groups in the race, he would meet with Demis right now and agree to slow down. But there is no cooperation or coordination between all the different groups involved, so no one can agree on anything. This, imo, is the main reason he wanted to restrict GPU sales: chip proliferation makes this kind of agreement impossible, and if there is no agreement, then he has to blitz. That seems to be exactly what he has decided to do. After watching his interviews today I think Anthropic is going to lean into recursive self-improvement, and go all out from here to the finish line. They have broken their cups, and are leaving all restraint behind them.

2025 = breakthrough research in AI for drug discovery; 2026 = the year of deployment at scale. Today we share that @ChaiDiscovery is partnering with @EliLillyandCo to accelerate biologics discovery.

The very first deal … with an industry leader … barely a year after founding … WOW! A huge testament to the incredible @chaidiscory team and the step change in research they unlock.

We’re thrilled to announce our collaboration with @EliLillyandCo. We’ll be deploying Chai’s AI models to power biologics discovery across a broad set of drug programs at Lilly. This is a major step forward in pharma’s adoption of AI-native R&D.