I wish I had a crystal ball for about 12 months from now in regards to stealing user data between Apple and Microsoft and I could make a dang decision between the two. I love just having a couple of devices for Windows for everything. But I hate this AI nonsense.
2 Likes
Well, if Apple sticks to its playbook and hordes your data for itself, then I’m in - better the devil you know than who knows what customer gets your data from M$ and/or Google…
Of course, this also depends how deeply Apple integrates with Google’s Ai - if it it the front end I think I’m good, but if it’s a hand off, CRAP!
4 Likes
Another article:
How a Windows shake-up could position Microsoft to capitalize on AI PCs - The Verge
1 Like
I understand enemy of my enemy and all, but I still can’t see why Apple would want to hitch its wagon to Google for Ai.
1 Like
The gang at MacBreak Weekly touched on it briefly the other day - competing directly with Gemini, Chat GPT, and Open AI will be very difficult in the short term, and Apple may be looking to “partner” with one or more as the front end for accessing those LLM’s - but I can’t see the LLM vendors wanting to do it (they lose the customer contact and data) EXCEPT for Google, who already pays billions to be Apple’s search provider. What if Apple opened that treasure chest to whomever gives them the best AI deal?
1 Like
Yeah they also mentioned on that episode that it might be for deniability in the case of hallucinations or other issues. They can just point to Google and “wash their hands” of the situation.
I think the “next race” will be to get AI to run locally. It sounds like with Nvidia that is more possible in Windows currently. If that happens, we could end up in a weird world where Apple has to send requests to Google and MS is running them locally?
1 Like
Except the GPU AI processing method is what Microsoft is now moving away from and strongly pushing to not use for local AI. Why? GPUs are incredibly wasteful and extremely taxing in energy use and are the equivalent of blandishing a sledgehammer to brute force AI. That is why Microsoft is requiring partner devices that want to be AI certified and AI compatible come equipped with an NPU with a minimum of 40 TOPs. So while NVIDIA may be riding high on the hog right now, the long view is that AI processing—much like what happened with cryptocurrency mining—is moving towards ASICs, specifically NPUs, which makes sense. Therefore, NVIDIA will need to develop ASICs or NPUs or their stocks will crater when their customer base sees just how grossly inefficient GPUs are for AI, and high tail it for the likes of Jim Keller’s Tenstorrent, or even in-house developed solutions like Microsoft’s own Athena AI chip.
2 Likes