This is a great resource, when I was in Saudi Arabia a few months ago at the global AI conference I gave a talk on AI in materials science - and while I was talking an LLM read a paper, created a hypothesis and used our API to create and run experiments - all live on our workbench behind me while I gave the talk.
LLMs aren't perfect by a long shot, but at the very least they can attempt to reproduce the paper, take a lot of the grunt work out of establishing a baseline, and let the scientist focus on making changes and pushing forwards with new experiments, rather than getting bogged down in replication of existing work.
