What area of AI/ML doesn't get enough energy?

Question

There&rsquo;s been so much focus on LLMs, image & video generation, and text to speech models lately.These models obvious have wide general purpose capabilities, so they are particularly useful to folks technical and non-technical.I&rsquo;m curious what (relatively) general model domains does the HN community not think get enough research and product building energy?For example do you think that image recoginition or non-verbal audio recognition isn&rsquo;t getting the direct sunlight it deserves from both developers and customers?

PaulHoule · Accepted Answer

People making their own training sets, defining problems that can be solved, rather than the dominant paradigm of &ldquo;prompt &lsquo;n pray&rdquo;.

bpiche · Answer

I&rsquo;m gonna go with interannotator agreement of labeled training data

What area of AI/ML doesn't get enough energy?

People making their own training sets, defining problems that can be solved, rather than the dominant paradigm of “prompt ‘n pray”.

I’m gonna go with interannotator agreement of labeled training data