Good taste matters more than ever in research

With AI, whether that's LLM-based chatbots or LLM-based agents, it's been easier than ever to ask AI "Suggest a novel research idea" and to "code this idea for me: [paste]".

This has only made the barrier of entry less for anyone interested in research and it leaves room for lots of AI research slop.

This is why, more recently, I've started believing that equally as important as your coding capability is "having a good taste for solving problems that matter."