prototype_g2@lemmy.ml to Fuck AI@lemmy.world · 4 days agoBonfire of the LLMs: Gartner forecasts ‘extinction’ for generative AI vendorspivot-to-ai.comexternal-linkmessage-square2fedilinkarrow-up125arrow-down10cross-posted to: techtakes@awful.systems
arrow-up125arrow-down1external-linkBonfire of the LLMs: Gartner forecasts ‘extinction’ for generative AI vendorspivot-to-ai.comprototype_g2@lemmy.ml to Fuck AI@lemmy.world · 4 days agomessage-square2fedilinkcross-posted to: techtakes@awful.systems
minus-squarealvvayson@lemmy.dbzer0.comlinkfedilinkarrow-up5·4 days agoI’ve said it often before. LLM’s are really great for many use cases. Transcribing, translating, summarizing, editing, code generation, etc. But the marginal difference between the cheap models like DeepSeek and the top-of-the-line is low. So take transcribing, translating, summarizing and editing. Why would anyone spend top dollar when DeepSeek does it basically free? The potential market for the state of the art models is very limited and will never recoup the 100s of billions being invested.
I’ve said it often before.
LLM’s are really great for many use cases. Transcribing, translating, summarizing, editing, code generation, etc.
But the marginal difference between the cheap models like DeepSeek and the top-of-the-line is low.
So take transcribing, translating, summarizing and editing. Why would anyone spend top dollar when DeepSeek does it basically free?
The potential market for the state of the art models is very limited and will never recoup the 100s of billions being invested.