• Moderators, please send me a PM if you are unable to access mod permissions. Thanks, Habsy.

Tech Thread

LeafGm said:
Is “Gemini” the AI that pops up at the top of Google search results?
Click to expand...

Yes, though I'm not sure which model of gemini it is.

LeafGm said:
If so, I don’t get it. Thing’s as dumb as a stump and completely fucking useless.
Click to expand...

That's where the model used is pretty important. Their newest (2.5 Pro) is more or less equivalent to GPT 4.5 in a lot of use cases.

Honestly not sure why that's Marques' take on this. Apple never really rolled out a functioning LLM model (they had a pretty famously bad roll out of "Apple Intelligence" not that long ago) so their only path was to pick somone's model to work with and Gemini is way cheaper to run than GPT. Also, and probably pretty importantly to a massive userbase like Apple is Google's scale. They own their own data centres (and most the data centres Open AI uses fwiw). Open AI legitimately might not exist in 3 years because they burn dump trucks full of money every day, don't own their own infrastructure, and their top model is 4-5x the cost to operate as Gemini 2.5 pro is, like ~15x what the scaled down version of 2.5 costs to run.

For a corp the size of Apple there were really only 2 companies worth building a relationship with on this. Google (Gemini), or Microsoft (Co Pilot). Everyone else are start up cash arsonists with no functioning business plan.
 
I mean it is weird that Apple is teaming up with Google (owner of Android) rather than developing their own thing. Especially concerning something as "Apple" as Siri.

I think the most recent version of Gemini has been well-received, so it may well be a good idea, especially if the cost difference (vs GPT) is that dramatic.
 
It's weird on the surface, sure, but only on the absolute surface. Apple has never taken model development seriously and aside from a bit of a fuck about with apple intelligence (which they clearly didn't invest a lot in), it's always been pretty clear that they were positioning themselves to be a user and not a developer, which will probably prove to be the smartest approach in the end with most of the developers bankrupt and the models that survive commodified.

Shit, Apple will probably be able to turn around in a few years and buy one of these burnt out tech startups with a decent model for a small fraction of what it would have cost for them to build it themselves.
 
  • Like
Reactions: CH1
That's the other problem for LLM AI corps. All of them are going to essentially end up identical with no business advantage to separate them (IMO).
 
You must log in or register to reply here.
Back
Top