monroewalker

Thank you for sharing, that provided some good context for how to interpret this

Post content:

_____

I wish we didn’t need these again, but here is the honest version of Anthropic’s biology announcement (Caveat: I haven’t worked in bioinformatics for many years.) The good: Anthropic ran ~950 Claude agents over a large biological sequence database. Claude searched, wrote code, compared sequences and genomic neighborhoods, and found an interesting pattern that apparently had not been noticed before: a known reverse transcriptase associated with another gene and a repetitive DNA array.

That is cool. Automating this kind of open-ended bioinformatics search at scale is useful, and Claude may have found a lead a human would have missed.

But: Claude did not do a biological experiment. It searched databases and analyzed data.

Humans then took the candidate into the wet lab. And the wet-lab result so far is modest: they showed that the repeat array produces short RNAs.

We still don’t know what the system does. No function, mechanism, phenotype, targeting, defense activity, or programmability has been demonstrated.

This is also where the CRISPR framing gets ahead of the result. Right now, “it has some features reminiscent of known programmable systems” is a hypothesis for what to investigate next, not a discovery that it behaves like CRISPR.

And there is a missing baseline: bioinformatics has had tools for finding unusual gene neighborhoods and candidate systems for years. The interesting comparison is 950 Claude agents vs. an expert using the best existing computational pipelines - not Claude vs. someone manually looking through 200,000 sequences.

So my honest announcement would be:

Claude autonomously found an interesting candidate for a previously uncharacterized biological system. A small human wet-lab experiment confirmed that part of the candidate is expressed. We don’t yet know what it does.

That is a good result.

But in a regular biology lab, this isn’t the finished paper. It is the result you show at lab meeting and say: “This looks interesting. Now we need to figure out what the hell it does.”

Maybe that next step leads to a major discovery. But that discovery hasn’t happened yet.

Seems more than good enough for a live model though! I can imagine this demo being extended to be a lot nicer to play with. You can just feed the model engine analysis and it can make as high of quality moves as needed. No longer any correlation between the model's understanding of the position and the moves that would be made but I think that's still a really nice improvement when thinking about this as adding live voice interaction to existing chess vs computer functionality rather than adding chess to possible interactions with the latest live voice model.

Same price as Fable?

Great summary! I would just add that cherry on top though -- that HuggingFace tried using the top commercial models in response but couldn't because of the cybersecurity restrictions so they had to use GLM 5.2 instead

"When we started the log analysis, we first used frontier models behind commercial APIs. This did not work: the analysis requires submitting large volumes of real attack commands, exploit payloads, and C2 artifacts, and these requests were blocked by the providers' safety guardrails, which cannot distinguish an incident responder from an attacker. We ran the forensic analysis instead on GLM 5.2, an open-weight model, on our own infrastructure. This had a second benefit: no attacker data, and none of the credentials it referenced, left our environment."

What makes it nicer to use than Slack?

Heh after reading that title card I thought this was going to be a mosquito based software bug analogy. I expected a description of how to write software that resulted in more "good bugs" that might facilitate finding other bugs somehow. Now I'm a little disappointed

Just about every line of code for this was written by Claude 3.7 via Claude Code. I never gave the other AI development tools like Cursor and Aider a fair shake so I may have just been behind the times on what's possible with agentic editors, but I found Claude Code to be extremely impressive. The API costs did end up being something around $15 though and this is a really small project, so I imagine cost goes up quite a bit to do any non-trivial work on a project of a decent size

This was great for me though because this is something I've had in mind for a while but never considered quite useful enough to be worth the time it'd take to learn about IntelliJ plugin development + the Java AWT knowledge required to create a canvas app like this

lol yeah I tried to get it to whisper too. And talk faster or slower or do accents. It seemed to be able to kind of do each of those things but only very slightly. Enough to see that there was some successful interpretation of the request but lack of flexibility to fully execute on it. OpenAI's model still has this beat on that front imo (talking quietly / slower / faster)

Yeah after a few interactions, the repetition of the mannerisms that initially added to the sense of life-likeness started to break the illusion a bit. The "you got me" response shows up a bit too often. The creativity remains impressive though