Michael Skarlinski (@m_skarlinski) 's Twitter Profile
Michael Skarlinski

@m_skarlinski

ML/Engineering enthusiast and Member of the Technical Staff @ FutureHouse

ID: 1787551300844232704

calendar_today06-05-2024 18:33:44

18 Tweet

179 Followers

5 Following

Sam Rodriques (@sgrodriques) 's Twitter Profile Photo

Today, we are launching the first publicly available AI Scientist, via the FutureHouse Platform. Our AI Scientist agents can perform a wide variety of scientific tasks better than humans. By chaining them together, we've already started to discover new biology really fast. With

Andrew White 🐦‍⬛ (@andrewwhite01) 's Twitter Profile Photo

The plan at FutureHouse has been to build scientific agents and use them to make novel discoveries. We’ve spent the last year researching the best way to make agents. We’ve made a ton of progress and now we’ve engineered them to be used at scale, by anyone. Today, we’re launching

The plan at FutureHouse has been to build scientific agents and use them to make novel discoveries. We’ve spent the last year researching the best way to make agents. We’ve made a ton of progress and now we’ve engineered them to be used at scale, by anyone. Today, we’re launching
Andrew White 🐦‍⬛ (@andrewwhite01) 's Twitter Profile Photo

Here's a nice concise demo of how you can use AI agents for quickly getting to the boundary of human knowledge and then proposing a novel hypothesis

Michael Skarlinski (@m_skarlinski) 's Twitter Profile Photo

A cool detail on our platform -- if you share a task after it finishes, a social media preview will be dynamically generated for each task. Surprisingly tricky feature to get right! Thanks to Tyler Nadolski for working this out. platform.futurehouse.org/trajectories/9…

Tom Ellis (@proftomellis) 's Twitter Profile Photo

I gave FutureHouse's platform a go today, challenging the Falcon agent to try to explain some perplexing counter-intuitive lab results we just got and suggest experiments to work them out. I have to say I was very impressed. I'd rate it at ~70% as effective as me... (1/3)

Sam Rodriques (@sgrodriques) 's Twitter Profile Photo

This is the most up-to-date comparison between Crow (on our platform) and recently released chat models with search on LitQA, a scientific retrieval benchmark (part of LAB-Bench). See paper in comments. Accuracy < Precision reflects refusal to answer.

This is the most up-to-date comparison between Crow (on our platform) and recently released chat models with search on LitQA, a scientific retrieval benchmark (part of LAB-Bench). See paper in comments. Accuracy &lt; Precision reflects refusal to answer.
Sam Rodriques (@sgrodriques) 's Twitter Profile Photo

Instructions on how to use FutureHouse Platform API. We got some feedback that this was unclear from our initial posts yesterday. Sorry about that!

Brian Naughton (@btnaughton) 's Twitter Profile Photo

We have gotten some really good responses to science questions from platform.futurehouse.org already. Both from "Crow" (short answers) and "Falcon" (deep research). It looks like this is state of the art right now!

The Information (@theinformation) 's Twitter Profile Photo

AI Agenda: The Startup Building an AI Scientist Why this startup building an AI scientist says we need a “Stargate” program for AI-driven scientific research. Read more from Stephanie Palazzolo 👇 theinformation.com/articles/start…

Sam Rodriques (@sgrodriques) 's Twitter Profile Photo

Introducing Finch, a new agent that fully automates data-driven discovery in biology. We are launching a closed beta for it today (sign up below). This is still early, but impressive, maybe similar to a good 1st yr grad student. In the video, see how it independently reproduces

Andrew White 🐦‍⬛ (@andrewwhite01) 's Twitter Profile Photo

We have an API for clinical trials on our platform - which means you can ask questions like "what trials will read out in June for NSCLC and how likely would you rate their success based on previous trials in the area." Pretty cool.

We have an API for clinical trials on our platform - which means you can ask questions like "what trials will read out in June for NSCLC and how likely would you rate their success based on previous trials in the area." Pretty cool.
Andrew White 🐦‍⬛ (@andrewwhite01) 's Twitter Profile Photo

FutureHouse's goal has been to automate scientific discovery. Now we used our agents to make a genuine discovery – a new treatment for one kind of blindness (dAMD). We had multiple cycles of hypotheses, experiments, and data analysis – including identify the mechanism of action.

FutureHouse's goal has been to automate scientific discovery. Now we used our agents to make a genuine discovery – a new treatment for one kind of blindness (dAMD). We had multiple cycles of hypotheses, experiments, and data analysis – including identify the mechanism of action.