The current generation of models is already being used effectively by scientists for research purposes.

The next generation models will need to understand physics and chemistry better, and while they may grasp Newtonian physics, discovering new chemistry or novel theoretical physics remains uncertain.

Current AI models feel limited and are not making groundbreaking discoveries like general relativity, possibly due to the way we frame our research questions.

The biggest barrier to increasing scientific progress with AI is that the models are just not smart enough yet.

The models are already at a point where the things they can't do don't seem fundamentally different from the things they can do, which makes hitting a wall less plausible now than before.

The model can interpret genetic mutations and provide insights into their functional consequences, which is state-of-the-art in computational biology.

The scientific community often struggles to integrate new discoveries into existing models, preferring to fit new information into pre-existing frameworks rather than allowing the data to guide the narrative.

The model learns from vast datasets of genomic sequences, allowing it to identify biologically meaningful patterns that can lead to new insights in biology.

We often underutilize models, expecting them to solve tasks quickly without giving them enough time to iterate and improve.