Thread

1. trebligdivad · 1790635054 · focus · HN ↗

   What proportion of commercial LLM use is classification? I'm just wondering what happens to business AI spending/data centre usage when they realise they don't need full LLMs.

   1. exogenousdata · 1790644061 · focus · HN ↗

      The American stock market probably loses 20% of its value in a few days.

      1. drstewart · 1790680389 · focus · HN ↗

         Can you share your short positions so we can see how confident you are in these predictions?
      2. causal · 1790690711 · focus · HN ↗

         Can you uhh elaborate on why?

         1. folienumero · 1790936860 · focus · HN ↗

            because entire economy is booming due to AI spending which priced in for more spending and revenue in the future.
   2. svachalek · 1790644779 · focus · HN ↗

      I'd expect the vast majority at this point is coding. Classification is a thing but in my experience tends to run on light, cheap models, not the proprietary frontier ones.

      1. RamblingCTO · 1790666874 · focus · HN ↗

         part of agentic engineering is classification as well tho. reviewing/gating, what to read etc. don't need a full LLM
   3. BowBun · 1790650263 · focus · HN ↗

      We've stopped upgrading the models of our classification workflows for >1 year at this point, meaning they're running acceptably on early/mid-2025 models. That said I believe there is a long tail of non-production-ized users who throw this into their everyday LLM chats.

      1. baobabKoodaa · 1790671787 · focus · HN ↗

         In one of my client projects we've had to do forced upgrades of LLM models because of deprecations (I believe three or four of them over the course of 2 years). Each time our internal benchmarks have shown REDUCED performance after upgrading to "better" models.

Unofficial Hacker News client; not affiliated with Y Combinator.