Spoke with a friend in the US regarding AI and LLMs.
-
Spoke with a friend in the US regarding AI and LLMs. He said that in the startup, entrepreneur circles if you're doing anything AI you're assumed to be using a frontier model, there's just no mental alternative.
I've gone the other route and even assuming that frontier models will disappear and we'll be left with smaller, "local", models in 2-3 years time.
I also find that frontier models are used for even the most trivial of tasks where cheaper, even publicly available models, will do the job just fine.
Personally I think that the next logical step is to see a more "intelligent" fabric dispatching LLM-type based jobs to where they are most appropriately performed. And that certainly doesn't mean one model does it all.
-
Spoke with a friend in the US regarding AI and LLMs. He said that in the startup, entrepreneur circles if you're doing anything AI you're assumed to be using a frontier model, there's just no mental alternative.
I've gone the other route and even assuming that frontier models will disappear and we'll be left with smaller, "local", models in 2-3 years time.
I also find that frontier models are used for even the most trivial of tasks where cheaper, even publicly available models, will do the job just fine.
Personally I think that the next logical step is to see a more "intelligent" fabric dispatching LLM-type based jobs to where they are most appropriately performed. And that certainly doesn't mean one model does it all.
@nopatience yup, i think so too.
-
R relay@relay.infosec.exchange shared this topic