3h ago

Isomorphic Labs' Bilal Khan argues AI labs must not reduce specialized scientists to training data annotators

Lu suggests ten-year research projects should remain in academia.

——0——
Original post
OPbilal#1402bilal@BILALTWOVEC

my two cents here are that i strongly agree with the direction we’re headed in but we should set our sights higher than prescribing scientists to all become training data annotators like the rest of us unskilled software engineers lol

4:10 PM · May 25, 2026 View on X
REPLYAmy Lu#1623Amy Lu@AMYXLU

My heuristic is that if area of research you're most excited about might not work for 10+ years, then go to academia where you don't need to be derailed to provide shareholder value. If your interest just so happen to be immediately economically useful -- and from your tweet this sounds to be the case -- then yes, capitalism is your friend, go to a privately funded research vehicle. And no matter where you stand on the RSI/ASI/etc. timelines, I think we should agree that there will still be *some* open problems for us (human and/or AI) 10 years from now

will depuewill depue@willdepue

academics are unprepared for the coming world where much scientific progress is majorly a function of inference compute. whether OpenAI points the Eye of Stargate at your particular field will decide its acceleration. talent will leach away into the labs. it's already begun

7:19 PM · May 24, 2026 · 355.3K Views
12:08 AM · May 26, 2026 · 82 Views
Isomorphic Labs' Bilal Khan argues AI labs must not reduce specialized scientists to training data annotators · Digg