Danish Kapoor
Danish Kapoor

Getting lost with AI has been Apple’s biggest potential mistake in years. He’s trying to fix it with money

I imagine Tim Cook in his office, in December 2022. “I’m going to try what I saw on TechGIndia from ChatGPT.” And then, increasingly amazed and angry, realizing something: they had gotten lost. How had it been possible?

Apple moves on from AI. In recent months, the market did not seem to be particularly concerned about an Apple that barely spoke (or speaks) about artificial intelligence. It is as if it completely passed on this promising technology, and that does not seem so terrible because Apple usually arrives late, but does so with products (iPod, iPhone) that surpass those of its rivals. However, at his last WWDC conference he did not utter those words even once, and although we think we know the possible reason why, the other big reason is simply that they had nothing to talk about.

We have gotten lost. As Mark Gurman points out at Bloomberg, Tim Cook may be saying publicly that Apple has been working in the field of generative AI for years. However, his sources indicate that the company was taken by surprise by these developments, and that “there is a lot of anxiety about this and it is considered a fairly big failure internally.”

Ajax. This is the name of the large language model (LLM) that the company has developed after starting to work in this area. It is the one used internally by their own chatbot, which has been dubbed “Apple GPT”, and they are now evaluating whether this development can compete with their rivals and how they will deploy it in their products.

The solution, spend money. John Giannandrea and Graig Federighi are responsible for these new efforts, although Eddy Cue, head of services, is also involved according to Bloomberg. To try not to lose that race, Apple is investing close to 1 billion a year.

Nothing especially new. Those responsible for Apple seem to be lagging behind the launches we have seen from other technology giants. Thus, the Cupertino teams would be working on a Siri with integrated generative AI (like ChatGPT with voice), autocompletion of sentences, integration of generative AI also in XCode (as in GitHub Copilot, among others) or an improvement of Apple Music that will generate playlists more in line with the user’s tastes (like the one Spotify presented in February). The important thing, at least for now, seems to be being able to compete with its major rivals with similar products but adapted to their ecosystem.

In the cloud or locally? At Apple, one of the big debates about how to deploy these services is whether they will depend on the cloud—as is the case with ChatGPT or Midjourney—or can they be used locally on the device even if we are offline. This last option has been frequently adopted by Apple and offers more guarantees in terms of privacy, but for now the cloud allows access to more advanced options.

Image | ACC District | OpenAI

Danish Kapoor