Omar Khattab

@lateinteraction

asst professor @MIT CSAIL @nlp_mit. ColBERT.ai, DSPy.ai (@DSPyOSS), GEPA, RLMs, Pedagogical RL

Cambridge, MA

Omar Khattab
@lateinteraction
Sep 15
and they delivered!! i hadn't expect the shape of this. congrats @CompleteSkeptic and team @typesafeai!
Diogo Almeida
@CompleteSkeptic
Sep 15
02:56
After co-inventing ChatGPT, I kept asking myself: why have superhuman chat models not led to AGI? I’ve spent the last 2 years in stealth building a new way to train models (RLCD), and a new type of frontier AI model that we are releasing today: Jev • 20-200x faster • 40-400x
9
Omar Khattab
@lateinteraction
Sep 15
oh i have a GREAT idea for how to fix this! first, we...
Omar Khattab
@lateinteraction
Sep 15
i find it a bit weird that, in Sept 2026, agents can't use tools or write to files during compaction - it's really just a summarization prompt
10
Omar Khattab
@lateinteraction
Sep 15
whatever you think of the labs' under-investment in clear and delightful writing, in favor of skills like math and coding, it's clear that it has cost them hugely in perception
11
Omar Khattab
@lateinteraction
Sep 15
i find it a bit weird that, in Sept 2026, agents can't use tools or write to files during compaction - it's really just a summarization prompt
61
Omar Khattab
@lateinteraction
Sep 13
my days are like `N` hours of working with frontier models, babysitting them in frustration and disbelief at the frequent errors and slop interrupted by `M` min breaks of browsing X takes about how the same models (and _especially_ the next gen!) are superhuman at ~everything
48