“The challenge of this argument that it's just like this, just a, uh, uh, uh size of dataset thing, is that currently even the state-of-the-art virtual cell models that exist saturate at a very, very small uh fraction of the input data that is fed to them.”

ConceptResearch

“even the state-of-the-art virtual cell models that exist saturate at a very, very small fraction uh of the input data that is fed to them”

ConceptResearch

“Prakash also made the case for specialists. A model like AlphaFold does its one job better than any generalist. He expects that to hold. And the version of superintelligence people fear is a single model that does every job better than the specialists and depends on no one. My...”

ConceptResearch

“It managed to discover a better auto-research bonus. Then we keep tuning in the last 2 years.”

ConceptResearch

“improving the efficiency of self-referential research is definitely already on the roadmap.”

ConceptResearch

“There's an interesting thing we observe on self-evolved uh auto-research harness. We see the code it generated.”

ConceptResearch

“The other thing that is borderline haunting to me, I just did an episode not too long with Bronson Shane from Apollo, and reading through some of these chain-of-thought sequences, when the models are referring to memories that they have developed from their training, that also...”

ConceptResearch

“Oddly enough, I think early image-video models weren't good at being human-realistic, right? You see a lot of the high-res 4K professional photography, but not just everyday life, like normal picture, right? Everything looks like it's professionally generated, like professiona...”

ConceptResearch

“I think the delta is also huge. I learned pics to pics, like this was intro to ML, the TensorFlow, like Jupyter, Google Colab notebooks were like this, and then you have a sudden step function change with diffusion and whatnot. Any other ones since that, like there were clear ...”

ConceptResearch

“we really believe that if we trained a video model at a large scale, uh we would get a great model at the end”

ConceptResearch

“I am shocked today by what we are seeing. And what's funny to me is these are not state-of-the-art models. We are not getting— I should say, we have got 2 brand new models this week. We got Opus 5.5. We got GPT-6 Sol and 6 Luna. We are going to get into what those mean. But th...”

ConceptResearch

“Again, amazing models out of OpenAI. These are really cool.”

ConceptResearch

“Astra is overall the kind of more useful model for at least the coding workflow type stuff”

ConceptResearch