Apple’s Google Gemini deal is the most disappointing thing to come out of Apple since the Newton
Apple will not be among the tech leaders in this golden age of AI. That much is now clear based on its stunning deal with Google to use Gemini Models as the basis for its Foundation Models and what will likely form the fundamental personalization and smarts of the next major Siri update.
I’m not arguing the efficacy of this choice. Google and Gemini are inarguably the best ones for Apple if it’s committed to letting a third party jump-start its model efforts. After all, though competitors, Apple and Google are also partners, with Google paying Apple a repoprted $20 billion a year for search integration on Safari.
A revise and then a retreat
With V1 not being up to “Apple Standards,” Apple made the difficult decision to postpone the personalized Siri rollout and switch to V2.
At the time, Federighi refused to name a delivery date, but he did say they’d wait “until we have in-house, the V2 architecture delivering…”
What Apple’s been able to build “in-house” is now questionable. Clearly, it’s not the necessary models or at least some key part of the models needed to build a Siri that understands you and your phone, perhaps better than you do.
I’ve been re-reading this part of the joint Apple/Google statement: “Apple and Google have entered into a multi-year collaboration under which the next generation of Apple Foundation Models will be based on Google’s Gemini models and cloud technology.”
I highlighted the key phrase, which I take to mean that Gemini models are the seed or foundation of Apple’s “Foundation Models.” Looked at another way, you don’t get Apple’s take on this key piece of AI architecture without Gemini. Note that Google’s “cloud technology” is included, which I think means Apple will be tapping into Google’s larger models for more difficult AI problems.
Taken a step further, that means Siri’s brains might be more Google than Apple. Granted, that might be taking things a bit far, since Siri will be on top of Apple’s Foundation Model (based on Gemini models), but in the absence of clarity and depth from Apple and Google, it’s hard to see this any other way.
For some, AI relief
In reading the comments from colleagues and consumers on social media, I’ve noticed many are actually pleased that things have turned out this way. Perhaps they got tired of waiting for a better Siri, certainly one that could stand on a level AI playing field with ChatGPT or Gemini.
I understand that sentiment. After all, millions of iPhone, Mac, and iPad users mostly care about the quality of their Apple ecosystem experience. A smarter Siri, even one powered by Gemini models, is sure to improve it.
Even some industry watchers like Creative Strategies President and Principal Analyst Carolina Milanesi agree that Apple’s approach, which appears to focus on the end-user experience and interface rather than on the underpinnings, might be best. “I really do think my F1 analogy is perfect: buy the engine and focus on the design of the car, and optimize for your driver, that is what will give you the best performance,” she wrote to me in a text.
I think that’s a generous take.
Understanding Apple’s original goals, though, may be instructive here. Apple’s never wanted to build a Gemini competitor. As Federighi told me last year, “This wasn’t about just building a chatbot. So, seemingly, when some of these Siri capabilities I mentioned didn’t show up, people were like, ‘What happened, Apple? I thought you were going to give us your chatbot. That was never the goal, and it remains not our primary goal.”
The plan was not to compete with OpenAI and Google on the pure-play prompt field, but that doesn’t release it from the responsibility of holding a position in the AI race.
Where is Apple in the AI race?
AI is unquestionably the most important technological innovation in a generation, and the idea that Apple can’t solve it, can’t build technology to compete is disappointing.
Apple is not necessarily taking the big swings it did 15, 20, or 25 years ago. It’s a business that’s growing based on how many services you subscribe to. It makes incremental updates to the iPhone, iPad, MacBook, and Apple Watch. These are excellent products that often have been so far ahead of the competition that there’s little need for radical change.
But attention has shifted from buzzy hardware and big new product categories to the power of the prompt and what AI can do for you. At a very fundamental level, Apple could’ve shown what AI could do for its customers by building the Siri it promised, the one that understands your phone and your needs and could take action on your behalf. And while it has effectively built a collection of AI tools in Apple Intelligence, it couldn’t build that Siri brain on its own, and now it’s handed the reins – at least in large part – to a chief AI competitor.
The good news is that with these new Apple Foundation Models quietly running Gemini models deep under the hood, we will finally see a Siri that, with your permission, can read your interface, see the apps you use, understand your data and the interplay of all of it to be proactive and a thousand times smarter when you say, “Hey, Siri.” But those gifts will not be solely of Apple’s making.
When Apple introduced the Newton personal digital assistant (no relation to today’s AI assistants) in the 1990s, it was so proud of the mobile note-taking device that it produced a lovely coffee book to go alongside it. The product failed, and all that was left was a book full of tech innovation dreams and ultimately empty promises.
This Apple, Google, Gemini, Siri deal is worse. We didn’t even get the existing product to play with. Instead, we got missed deadlines and a tacit admission that Apple wasn’t up to the task.
Follow TechRadar on Google News and add us as a preferred source to get our expert news, reviews, and opinion in your feeds. Make sure to click the Follow button!
And of course you can also follow TechRadar on TikTok for news, reviews, unboxings in video form, and get regular updates from us on WhatsApp too.
