I agree that the closer an LLM’s language matches its actual process, the more introspective it seems. I worry that if we put them into even semi air-gapped robots that doubt/confidence might emerge locally from explore/exploit dynamics and trap them in endless cycles of anxiety or even anguish.

I think you’d need to work with a lab that uses the Olink Explore panel and it would run between $500 and $1,000 per sample. If you can get the data in the right format, the R code is on GitHub.

Maybe more like a meme?

Discovery is impossible if funding is limited by public comprehension.

Worship not the brand.

Your work on thermodynamics has caught my attention.

Yes, but how coarse do you like your grains?

Methodologically innovative at least.

I used to love a good em dash and a negation framing (it’s not X—it’s Y), but now they trip my bullshit sensor.

Strip away the background and the foreground disappears.

Good news.