How Apple Could Help With AI and LLMs
A modest proposal for how Apple can come to dominate the AI race without ever actually matching with the current leaders[…]
[…]
So what is Apple the platform vendor to do? Stop thinking like an app ecosystem participant or even an app store referee, and start thinking like a platform vendor again. Apple’s value-add here will never be by producing its own “frontier” model that does everything best.
[…]
Right now all these AI providers are returning documents in response to user intent, and everyone is treating those documents like the finished product. Here’s your poem. Here’s your picture. Anything else?.
The brilliance of OpenDoc was that each of those were raw material.
Build a semantic index (SI), and allow apps to access it via permissions given similar to what we do for Address Book or Photos.
[…]
And similar to the Spotlight indexing API, developers should be able to provide data to the SI along with rich metadata.
[…]
But give developers the opportunity, and then customers will have something to choose from. Make the Mac and iOS the best platform to build personalized LLMs.
Let the apps die and live based on their own merit and reputation. Apple can build the platform, and maybe expand on it over time and use it themselves.
Apple Intelligence has so far been the opposite of the platform strategy that had previously served it so well. Modern Apple is more about providing closed solutions and services.
Previously:
Update (2025-03-25): John Gruber:
But Apple didn’t make any of the actual apps. Companies like Adobe and Macromedia and Aldus did. Independent small developers made niche extensions for use inside apps like Photoshop, FreeHand, and QuarkXPress. When a new app came along like InDesign — which quickly ate Quark’s lunch — the Mac remained the dominant platform to use.
Making a great platform where other developers can innovate is one of Apple’s core strengths.
[…]
Apple should be laser focused on doing this for AI now. Where I quibble with Mueller is that I don’t want Apple to get out of the way. I want Apple to pave the roads to create the way. Apple doesn’t have to make the cars (literally) — just pave the best roads. Make the Mac the best platform for outside developers to create innovative AI systems and experiences. Make iOS the best consumer device to use AI apps from any outside developer. Work on APIs and frameworks for the AI age.
Do it like Health — a model that has proven to be:
- profitable (for Apple itself, selling devices like Watches);
- popular (with users, who actually use it, understand it, and like it);
- private;
- and open to third-party developers, device makers, and medical service providers.
One of Apple’s specialties is in improving the experience of using many of the same technologies as everyone else. I would like to see that in A.I., too, but I have been disappointed by its lacklustre efforts so far. Even long-running projects where it has had time to learn and grow have not paid off, as anyone can see in Siri’s legacy.
What if you could replace these features? What if Apple’s operating systems were great platforms by which users could try third-party A.I. services and find the ones that fit them best? What if Apple could provide certain privacy promises, too? I bet users would want to try alternatives in a heartbeat. Apple ought to welcome the challenge.