Ok check this out and the biochemist...
Starting at 06:28
6 minutes, 28 seconds
I'm trying to solve the scientific reproducibility crisis with an agent swarm. So over 50% of science can't be reproduced so they have a result that
6:36
6 minutes, 36 seconds
nobody else can replicate. And so I'm using a team of agents that essentially takes a paper, writes a uh extracts the
6:43
6 minutes, 43 seconds
methodology, and then writes a script for a lab automation robot called OpenRS to try all of the liquids and see if they actually get the same results. And
6:51
6 minutes, 51 seconds
then I also have a dry lab mode which will just run uh a GitHub repo or library if it's linked to the article to see if it gets the same figures and if
6:59
6 minutes, 59 seconds
even just like the dependencies install correctly. So focus on chemistry or like what's the use case? More like biomedical. Biome.
7:05
7 minutes, 5 seconds
Yeah. Yeah. And I'm using Daytona for all of the scripts and Tavali to scrape the articles.
7:09
7 minutes, 9 seconds
That's incredible, right? Like what's your background? Like it's a very unique, you know, niche thing.
I'm trying to solve the scientific reproducibility crisis with an agent swarm. So over 50% of science can't be reproduced so they have a result that
6:36
6 minutes, 36 seconds
nobody else can replicate. And so I'm using a team of agents that essentially takes a paper, writes a uh extracts the
6:43
6 minutes, 43 seconds
methodology, and then writes a script for a lab automation robot called OpenRS to try all of the liquids and see if they actually get the same results. And
6:51
6 minutes, 51 seconds
then I also have a dry lab mode which will just run uh a GitHub repo or library if it's linked to the article to see if it gets the same figures and if
6:59
6 minutes, 59 seconds
even just like the dependencies install correctly. So focus on chemistry or like what's the use case? More like biomedical. Biome.
7:05
7 minutes, 5 seconds
Yeah. Yeah. And I'm using Daytona for all of the scripts and Tavali to scrape the articles.
7:09
7 minutes, 9 seconds
That's incredible, right? Like what's your background? Like it's a very unique, you know, niche thing.
7 minutes, 13 seconds
Yeah,
definitely more science. I spent a lot of time in academia, so I've
seen this problem firsthand. I mean, we'd spend months literally doing
something just to notice that the paper we were replicating didn't even
work.
You are explaining me Chummer, that you build a robot exchanging the laboratory staff to tell you by trial and error that the base paper is bad. Just build a pipe to an online AI and let it check the paper. So, do you see the little paperclip icon? Try that and type afterwards: Please check the paper for scientific viability.
Aka. Why we have no jobs... Hackers hereby referred to as Deggers.
That comes from, I quote:"...then writes a script for a lab automation robot called OpenRS" and "...just to notice that the paper we were replicating didn't even work." Meaning not the procedure, but the instructions are the error source.
Do you want me to spell "viability" for you, still?
#cyberpunkcoltoure
