Image created with gemini-2.5-flash-image with claude-sonnet-4-5. Image prompt: Cinematic shot of silhouetted scientist at laboratory window, petri dish with bioluminescent blue cultures glowing in sharp foreground, massive particle accelerator ring illuminated in background through glass at dusk, Fincher teal-orange color grade, 35mm film grain, shallow depth of field, desaturated with crushed blacks and warm sodium vapor highlights, widescreen composition
Recently, LLMs were found to encode different languages in similar ways, a sort of Platonic representation of words. It now extends to science:: 60 ML models for molecules, materials & proteins (all with different training) converge toward similar encoding of molecular structure”” https://x.com/emollick/status/2005991284511760421
What does it tell us about human intelligence that half of yeast’s essential genes are interchangeable with ours? The genome encodes hyperparameters and reward functions, not learned weights. And according to @AdamMarblestone and @steve47285, evolution put it’s cleverness into”” https://x.com/dwarkesh_sp/status/2006170099699573113
.@AdamMarblestone explains the value of all these efforts to have AI automate formal math reasoning: “”I think of it like when everybody had to write assembly code. The amount of cool startups that got created was a lot less. Fewer people could do it. Progress was more grinding,”” https://x.com/dwarkesh_sp/status/2006454058148921448
Leave a Reply