News for hackers. Stuff that matters.
Re:Genie 2: A large-scale foundation world model by int_19h on Wednesday December 04, 2024 @02:55PM (#42321377)
Apple use of AI is based on their own models first, and they have spent an inordinate amount of effort on privacy. They fall back onto OpenAI / GPT for stuff that they know their models can't do well, which is not surprising because, like it or not, OpenAI still has the best models on the market - so it makes perfect sense to use them for tough-to-crack stuff (it's certainly better than spectacularly failing like e.g. Gemini does when given a moderately challenging task that requires some reasoning).

Google's AI offerings are in a very weird spot right now. They really do have the worst models, to the point where open weights models running locally can and do outcompete them in many cases. Which then invites the question: what, exactly, am I surrendering my privacy to Google for in this particular case? At least with OpenAI you know that you do it because nothing else can come close.

Re:Genie 2: A large-scale foundation world model by talldayo on Wednesday December 04, 2024 @03:27PM (#42321696)
Privacy is something you give up for all online software. I give up privacy to use YouTube, search engines, Hacker News and even DNS. Unless you can show me the source code for Apple's implementation, I daresay they're forcing you to give up a modicum of privacy too. Same goes for OpenAI's servers, for that matter.

Mostly I dog on Apple and Microsoft because both companies didn't even attempt to compete. They ignored machine learning when it was taking off (FAIR/BERT era), refused to ship complex GPU designs instead of NPUs though it was the only thing working (eg. Nvidia vs everyone else), and they continue to lag behind all their peers in local model development. The most impressive contribution either company makes to Open Source AI is the support they write for Tensorflow and Pytorch - neither of which they even own. If you erased every contribution that Apple and Microsoft made to AI research, the entire field could pick up where it left off tomorrow.

Google's got problems, but AI is something they are far-and-away better at than their contemporaries. It is just Google and Facebook at this echelon, which is embarrassing but frankly expected given how stagnant and anticompetitive Apple and Microsoft's business models have become.

Re:Genie 2: A large-scale foundation world model by int_19h on Wednesday December 04, 2024 @07:52PM (#42323695)
Apple Intelligence runs on-device, mostly. When it needs to do anything in the cloud, it tells you. And if you looked into how they have set things up, it's a much more private arrangement even in the cloud, compared to the others.

And of course there's no privacy with OpenAI either, not anymore so than with Google. But at least I know what I'm buying with that - an actual SOTA model, not a runner up. Google can't even beat Anthropic.

Re:Genie 2: A large-scale foundation world model by luckydata on Wednesday December 04, 2024 @03:27PM (#42321692)
If you work from false premises you can get to whatever conclusion you like. Gemini is not bad at all but you do you.
Re:Genie 2: A large-scale foundation world model by int_19h on Wednesday December 04, 2024 @07:54PM (#42323708)
"Bad" is a meaningless subjective assessment. The important question is whether Gemini is better than GPT-4o and GPT-o1. In most cases, it is not, and it's very noticeable if you do more than summarization or translation. Google is firmly behind Anthropic even, never mind OpenAI.