A paper published just a few days ago reported on the way to not only automate AI research but to also make it scalable, the more compute power you give it the smarter the AI it can make.
From the abstract in the above paper:
"This is a fully autonomous system that shatters fundamental
constraints by enabling AI to conduct its own architectural innovation. It can conduct end-to-end
scientific research in the challenging domain of architecture discovery, autonomously hypothesizing novel
architectural concepts, implementing them as executable code, training and empirically validating their
performance through rigorous experimentation, [...] we establish the first empirical scaling law for
scientific discovery itself demonstrating that architectural breakthroughs can be scaled computationally,
transforming research progress from a human-limited to a computation-scalable process."
There is also a pretty good video about this:
John K Clark See what's on my new list at Extropolis 3ed