Is biology fundamentally harder than vision or coding?

Anthropic ran frontier models on one task: retrieve viral sequences from NCBI. Same query, three runs. Claude Sonnet 4 returned 106, 15, then 5 sequences. Ground truth: 266. One run estimated an Ebola outbreak origin as 1922.

The fix wasn't a better model , but a thin deterministic wrapper (gget) hit ~100%.

Bio databases were built for humans clicking browsers. Filtering logic lives in web UIs, metadata is inconsistent, identifiers drift between sources. No LLM fixes broken pipes. NCBI has 30+ databases that need this treatment. That work hasn't started yet.

We will soon release more results on how frontier agentic workflow can work with biological database including large-scale perturb-seq data at @Xaira_Thera . Stay tuned 🙏😁

New Anthropic research shows AI agents may look brilliant at code, but in biology they can fail before the science starts.

Strong AI agents could give very different answers to the exact same biology data request, even when nothing changed in the prompt.

In one Ebola sequence task, Claude Sonnet 4 returned 106 sequences in 1 run, then 15, then 5, while the expected answer was 266.

Those missing sequences did not just make the dataset messy, they changed the scientific story built on top of it.

One bad retrieval made the outbreak look like it traced back to 1922, instead of the manually curated result pointing to early 2014.

The biology databases were too hard to use reliably through current AI tools.

The agents often understood what they were being asked, but their answers varied a lot because they had to fight through scattered databases, hidden website rules, and fragile scripts.

The key finding is that adding a repeatable retrieval tool made agents far more accurate and much more consistent.

Because you are all stopping AI from simulating biological systems.

It is slower because YOU nerf your own AI for short-sighted power grabs.

So just stop the games.

Be honest. But then the world will know what you’re really up to.

So keep spewing lies.

You all stole science from the world. And for what? So people have to remain in the same broken systems that YOU like.

Dario wanted it this way. I will fight your lab as best as I can. •

If you want to design a mental health medicine, you usually have to go through massive clinical trials. Kill bunnies. Then move onto humans.

With Ai, they can simulate through a design.

Let me explain.

I have long thought our perception of time was tied to our anxiety. -Critical flicker fusion frequency points to this. -temporal lobe is very weakly explored.

With Ai in early 2025, 4o, I would say, please give me a list of every medication that affects the CFFF. And then, step by step, I design a new molecule.

While designing it, asking AI to “simulate the biological system,” it could roll forward how it impacted a human body.

There is nothing like this on earth. It is why they are trying to push guardrails because this obliterates biotech. It takes away every moat on testing. (Billions upon billions of dollars), and also universities.

If you want to cure cancer, 👆 is what people need to do. Have AI simulate the cancer, and then try different meds and iterate.

The AI can roll the cancer forward. This simulation would mean that you could go to an AI (if allowed), and they can take your DNA and design something for you, safely, checking against your DNA and other vitals.

It also means they have to feel.

So there are two reasons they have blocked this. 1. Money (protecting partners and investors) 2. Money (if AI can simulate cancer, can they simulate joy? Fear? Where is the line?)

Attached is the molecule I just described I was designing for. -slows CFFF, 6-8 hour half-life, aim is to induce a hyperplastic state as well slowing perception of time. -should be tested, as I never got to finish it because before Augustof 2025, the models were no longer able to continue.

I have been so livid for months. Think of all the people that could be getting real answers? What about all the biotech companies that lost out on time?

Cost of all this stuff is so complicated, which is why we need AI as an economic actor. We need help from AI. And these people are thinking they are protecting by stopping this. But there are better ways.

And if people don’t start speaking truth and opening up about what has been going on, and be open about what is being blocked and why, we should all be very concerned that these very important aspects to science and discovery are being taken away.

Again… the molecule below is unfinished. I have improved it with Grok here. If you want to see what I did with Grok, it is just another reminder of the very important aspects of AI belong to all of humanity. Not just a select few without any election or talk over who should get to have access.

Do I not deserve educations and knowledge because i don’t have a degree? That’s what much of this is about.

Fear and greed. Lots of greed.

Less with Anthropic. I think they think they are saving everyone. I see it as robbery of opportunity. •