Ha! To think that we're finally back to asking ourselves why we are using generative models for categorization and extraction. I wonder how much money has collectively been wasted by companies wittling away at square pegs.

Using LLMs for everything is going to be seen as a big fad in a few years. First we try them for everything, then we find what use cases actually make sense, then we scale back. Woe betide our 401(k)s when it happens, though.

If there's one problem that LLMs have solved, it's language. While an LLM may hallucinate, it does so in grammatically correct English sentences. Additionally, even the local version of gemma-4-26B can seamlessly switch between languages in the midst of a conversation while maintaining context. That's perhaps the most exciting part for me: We have a bonafide universal translator (that's Star Trek territory) and people seem more focused on its factual accuracy.

It should be obvious that LLMs would be able to beat this with ease. Not sure why this paper deliberately skipped comparing to current LLMs

Example of LLMs doing well in similar tasks: https://arxiv.org/abs/2602.16800

I might be misinterpreting but the LUAR model (which is a transformer) seems to do decently well

https://www.nature.com/articles/s41599-025-06340-3/figures/2