4 Comments
User's avatar
Konrad Kording's avatar

Great writeup. I am equally struggling with this space. I think my personal answer is that hybrids, where a human works on the core idea while AI does the mechanical stuff around them is most promising.

Try my version at planyourscience.com

Konrad

Ubadah Sabbagh's avatar

I agree with you, especially in the near-term. Providing the scientist with more bandwidth (by doing mechanical work for them) and range (enabling them to more readily explore new fields) is an immediate value-add that can happen now.

Tried out the website, love the idea and love the "terms of use".

Rockwell Anyoha's avatar

Incredible read, so many well-synthesized ideas! The idea of scaling across modalities (I'd even lump the instrument-first idea here) is particularly eye opening (rather than do it again, do it a different way). Also love the very practical outline of how these ideas can be implemented immediatley. Also also, thanks for providing all these sources, I have a lot of supplemental reading added to the queue!

Ubadah Sabbagh's avatar

Thanks for reading!