Apple Is Losing Its Brains: Why Siri Isn't Getting Smarter (and Who Is to Blame)

Secrecy has always been Apple's core superpower, but in the age of generative AI it has suddenly become its main weakness. While Mark Zuckerberg hands out the weights of his Llama models to anyone who wants them, and Google DeepMind turns out research papers one after another, Cupertino keeps playing the silent game. The result was predictable: the best AI specialists have started packing their bags. Over the past few weeks, Apple has lost at least four key researchers and one of the top executives responsible for developing Siri. The irony is that they are not leaving for obscure startups but for direct competitors: Meta and Google DeepMind.

This situation has been building for a long time. In the AI world, a researcher's reputation rests on publications, conference appearances and contributions to open source. Apple, by contrast, traditionally demands complete anonymity from employees and forbids sharing work before a product's official release. For an ambitious engineer who wants to push the industry forward, working at Apple today looks like a golden cage. You can build a breakthrough technology, but nobody will know about it until Tim Cook shows it on stage three years later. With the AI field changing every week, such slowness looks like professional suicide.

The departure of a leader from the Siri team is a separate warning sign. Apple's voice assistant has long been a punchline for its limitations compared with ChatGPT or even the updated Google Assistant. Inside the company, there is clearly a fight over exactly how to integrate LLMs into the iOS ecosystem. Losing key people at this moment means that ambitious plans to "revive" Siri by the next WWDC could be at risk. Apple is now in the position of a chaser, and chasing without a strong engineering team is a nearly impossible task.

Competitors are acting aggressively. Meta offers researchers not only huge salaries but also access to colossal computing power and, above all, freedom of action. Google DeepMind, despite its bureaucracy, remains a Mecca for those who want to do fundamental science. Against this backdrop, Apple looks like a conservative bank trying to hire rock stars: the terms are good, but playing music is forbidden. If the company does not rethink its approach to secrecy and give engineers more freedom, the brain drain will turn into a full-scale exodus.

Interestingly, Apple has already tried to loosen its rules by allowing employees to publish some machine learning papers, but judging by the latest news, that has not been enough. Researchers want to work where the future is being made in real time, not where it is packaged into a pretty aluminum case with a two-year delay. Apple now faces a hard choice: either stay true to its principles of secrecy and lose the AI race, or open up to the world and risk its magic for the sake of survival.

Bottom line: Apple will have to either radically change its culture of secrecy or accept the role of a company that buys other people's AI technology instead of building its own.

*Meta is designated an extremist organization and banned in Russia.