Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
Apple Loses Another Key AI Leader to Meta's Superintelligence Labs
- Thread starter MacRumors
- Start date
- Sort by reaction score
There’s a very big gap between using AI to create generic image slop or an annoyingly chatbot and any supposed AGI.
I don’t think either Apple cares much at all about their “AI” placing, or that AI has had any significant impact on most people’s lives.
If a when the latter becomes relevant, the former probably will too. But not before.
Or they also might figure this AI money spending spree can’t last with no real path to a ROI. All these companies are spending billions. And I feel like it is going to backfire. If Apple gets its act together and delivers what it showed. I could come out ahead. Truly AI for the rest of us.
All company values are based on speculation. Do you really think that Tesla was ever worth more than Toyota and Volkswagen and Ford combined?
AI is still in an R&D phase. We're essentially beta testing and training the models. LLMs aren't going to make a magical leap to AGI, but they're good at task-based productivity improvements when people know how to use them. It's not a mature product.
If these models weren't available to the public, all of these companies moving toward AI development would otherwise be spending billions anyway with $0 of income and 0% ROI. AND they would have run out of training data. Every user that uses them is more data to train the models. Every time they're corrected by a user, it's more data to help improve error correction (the most expensive and energy intensive part of AI computing).
The LLMs will get better, they will get more efficient and they will eventually hit a wall because language is an inherently flawed way to process information and describe the world. But they will be combined with Open-World testing (real or virtual) which will allow them to learn about logic, physics and mathematics on their own. Since a lot of those processes are probability based, then that will finally be the real-world application that Quantum computing doesn't currently have - at which point, processing will become faster and cheaper.
LLMs aren't the destination. They're the human-interface overlay onto the next, deeper level of AI processing which is currently in development.
It's still a fair way off, but those two fields will eventually merge into one. It's why Google has Deep Mind testing open world simulations and they're investing heavily into Quantum computing development.
100% agree if it were just about $$$. Apple could buy every lab if they wanted if it were just about $$$. I suspect there are lots of poison pill clauses in the strategic investments M$, Google and Amazon put into the likes of Antrhopic and OpenAI.
Six of my friends have interests on VR/AR/MR things. Only one of them uses Vive as primary VR headset, others use Quest 2 or 3. Though I know they all bought Quest as second hand.