Zéro spam. Nous ne partageons jamais votre adresse, et vous pouvez vous désabonner à tout moment. Déjà abonné ? Connectez-vous
Professor of Machine Learning at the University of Oxford
"The AI build-out... is increasingly financed not by selling overpriced equity but by contracting future cash flows and borrowing against hardware.... That is not the architecture of 2000. That is the architecture of 2008." www.groundbrkr.com/p/the-second...
A hill I will die on: "Foundation Model", "Large Language Model", "Reasoning Model", "Generative AI", and even "Deep Learning" are all bad names. The field keeps naming things by what it hopes they are rather than what they do.
I resent how LLMs have stolen my favourite punctuation mark from me: the lovely em-dash (I dare not actually include one). And they don’t even use it right!!
I’ll be honest, a lot of these “AI will never be as smart as a human” takes read like they were written by a human
AI has decisively lost trans women, surely one of the most tech-positive demographics around. Trans women will happily spend hours discussing their favourite Linux distro, but do not admit to finding LLMs useful in a trans discord. Shame on Big Tech for draining the joy from my special interest.
As true for AI as anything else: racing only makes sense if you know there is a finish line and it is a place you want to reach
Is modern AI safe? No. But is modern AI at least becoming more safe? Not really. But is this powerful technology in the hands of folks who have repeatedly proved themselves to be competent, wise and motivated only by the greater good? Errr
Center = Michael Osborne. Left = members they follow (green edges). Right = members who follow them (blue edges). Top = mutual follows (orange edges, slightly larger). Drag any node to reposition; click to open that profile.
