Latest posts 5
Yesterday I was in the audience for what I believe is the first public unveiling of Prometheus from @microsoft. Immortal words - "compared to Prometheus, GPT3 is a toy." All I can say is, I agree. Certainly re clinical diagnosis to Talmudic scholarship which is what I saw - wow.
@sanjeevanahilan re #2. #solveintelligence
My unwavering opinion on current (auto-regressive) LLMsView quoted post on X
1. They are useful as writing aids.
2. They are "reactive" & don't plan nor reason.
3. They make stuff up or retrieve stuff approximately.
4. That can be mitigated but not fixed by human feedback.
5. Better systems will come
Thanks so much for your leadership and vision in putting this together @drjbajwa
Sincere thanks to June Raine @MHRAgovuk @basslinetherapy @HarvardChanSPH, @dramyabernethy @Verily for joining me on this discussion- hope this is beneficial to the wider community of #innovators #regulators #academics, so that together we can truly unlock the power of AI4healthView quoted post on X
Impressive work with huge potential clinical impact from @david_sontag and his lab re LLMs applied to a clinical corpus. New state of art performance in 4 areas WITHOUT retraining (zero vs few shot). Compelling both from pt and clinician workflow pov. arxiv.org/abs/2205.12689