How Anthropic's new results post would read without the PR:
Claude orchestrated open-source protein design models, PXDesign, RFdiffusion, Genie, BoltzGen, from a 30k-token expert prompt and 12,500 H100-hours of compute, and designed binders against 14 of 15 targets. Hit rates of 22–35% against a 10–15% baseline, where some of those tools already report similar numbers on their own.
The orchestration is genuinely impressive. But the open-source models did most of the lifting, and they came from the Baker lab, Columbia, MIT, ByteDance Seed, and most of them were already wet-lab validated before Claude touched them.
Which also sets the ceiling. All these generators share a single PDB-shaped training distribution, so calling four of them doesn't diversify away the blind spot, since they fail together. The targets that worked are the well-studied ones.
So the valid claim is that an agent can now drive this stack competently in the regime where the stack already works.
Instead, we got this announcement:
@VPrasadMDMPH You accuse some doctors of lacking critical thinking, then give patients simplistic heuristics that actively encourage them to secondguess evidence-based treatment. This is some top tier misinformation.
@VPrasadMDMPH #2 is genuinely dangerous patient advice. Confuses absence of symptoms with absence of biological risk. Metastatic cancer is not managed by waiting for the patient to feel sick. biology, burden, subtype, disease-specific evidence ect determine urgency, not night sweats.
@VPrasadMDMPH And #5 is almost comically reductive. Judging a novel clinical trial by “is the ORR >10–20%?” First in human trials literally exist because the efficacy is unknown. ORR also tells you nothing by itself about PFS/EFS/OS, QoL, toxicity, CNS activity or biomarker-defined benefit.
Most patients don't realize their doctors and nurses think about them in the spaces between visits. While driving home, getting ready for bed, while falling asleep.
Our work is a part of us, and we care about you. That's all.
@burkov We need voices like yours to continue exposing these grifters. Your previous efforts actually made them dial back their AGI/ASI hype for a while. Now it seems they've simply latched onto a new engagement engine to exploit
@littmath His intellectual dishonesty is obvious. He serves as a sycophant for OpenAI and is a prominent member of the dogmatic e/acc faction. They've abandoned rational inquiry, treating AI not as a technology but as a belief system to be defended with religious fervor.
@GaryMarcus@DeryaTR_ He won't critique OpenAI because he's in bed with them. AI influencers love to oversell their machine god. In oncology, we already have deep learning models that actually help us with risk stratification and prediction, and that's already very beneficial. Some examples below.
@burkov What pisses me off most is his 'oops, didn't expect this' act. Now that his erotica tweet blew up in his face, and boom, back to pretending he never saw it coming in his latest tweet. Everyone roasts Elon, but this guy can't be honest for one second.
@GaryMarcus@WIRED@StevenLevy He's back to hyping things up, just said a couple days ago that 2025 is the year AI surpassed human intelligence. DevDay is Monday and expectations are sky high. Whispers of the IMO/ICPC model dropping as an o3 type shockwave to silence the critics.
@sama They don't call you Scam Altman for nothing. The disingenuous "oops, didn't expect this much usage, GPUs melting" excuse just like the Ghibli theft situation is disgusting. I hope artists and IP holders fight back against this daylight robbery.
@burkov@xai For my personal use in medicine and oncology, this is not on par with Gemini 2.5 Pro or even R1. However, it is extremely fast, so I believe it can be useful for simple tasks or everyday queries that don't require complex reasoning. They traded depth for speed.
@LSXYZ9 People like you, who cashed out on Saudi money and/or turned a blind eye to Riot’s treatment of female workers, don’t get to act as moral arbiters of any kind. Bwipo’s comment is moronic but passing judgment while ignoring your own shadow is the height of being a gross hypocrite.