so much for recursive self improvement, to the degree that it requires scientific taste

so much for recursive self improvement, to the degree that it requires scientific taste the other thing im noticing while working on my research projects is how limited these models are GPT-5.5-xhigh is not even close to being an automated researcher you should also not ask these models for advice on how to design experiments. the models have 0 taste

Source: Gary Marcus (X) | 2026-07-06