Apple’s AI Struggles Revealed
What went wrong with Siri? Apple had big plans to make Siri smarter. But a recent report by Bloomberg’s Mark Gurman shows things didn’t go as planned. Behind the scenes, Apple has faced internal delays, leadership missteps, and rising pressure to keep up in the rapidly evolving world of AI.
The ChatGPT Effect
Everything changed in 2022 when OpenAI launched ChatGPT. Apple hadn’t planned any major AI features until then. But when Craig Federighi, Apple’s software chief, tested ChatGPT to write code for a personal project, he realized its true power.
That experience pushed Federighi—and Apple’s leadership team—into action. They began meeting with top AI companies like OpenAI and Anthropic. Soon after, Federighi instructed his team to pack iOS 18 with as many AI-powered features as possible.
Falling Behind the Competition
John Giannandrea, Apple’s former AI chief, formed a new team to develop large language models (LLMs). However, early tests showed disappointing results. Apple’s chatbot was about 25% less accurate than ChatGPT.
Under pressure and running out of time, Apple decided to seek outside help. Giannandrea reportedly preferred a partnership with Google, valuing its stronger privacy protections. But Apple ultimately chose OpenAI and confirmed the deal during WWDC 2024.
One major upgrade promised was the ability for Siri to send difficult questions to ChatGPT. That feature is now delayed until December. Other AI features are also missing, sparking user complaints and even lawsuits.
Internal Confusion and Technical Hurdles
Even within Apple, the rollout hit major roadblocks. Federighi, testing AI features on his own iPhone, found them too buggy before iOS 18.4. Meanwhile, Apple’s marketing team promoted features that weren’t ready. Ads had to be pulled shortly after they went live.
Much of the blame fell on Giannandrea. Some employees felt he moved too slowly. Others believed he didn’t trust generative AI enough to go mainstream. One executive said he doubted the technology’s readiness and felt marketing had oversold its potential.
Eventually, CEO Tim Cook replaced Giannandrea with Mike Rockwell, who previously led development on the Vision Pro.
Siri’s Code Problem
Siri’s aging architecture also held Apple back. The assistant’s code wasn’t built for today’s AI demands. To make upgrades, engineers had to split Siri’s system in two. Even simple features—like setting an alarm—became difficult to implement. Merging the new AI engine with the old system caused even more bugs.
Now, Apple is rebuilding Siri from scratch.
A New Start for Siri
After years of internal challenges, Apple is finally addressing what went wrong with Siri. A specialized team in Zurich is working on “LLM Siri,” a fully redesigned version powered by advanced language models. This new Siri aims to be more natural, responsive, and intelligent in everyday conversations.
In a surprising move, Apple is also preparing to let users in Europe replace Siri with third-party voice assistants. This marks a major shift for a company long known for its tightly controlled ecosystem—and a clear sign that Apple is taking steps to correct what went wrong with Siri.
What’s Next?
According to The New York Times, some of Siri’s new features may still arrive this fall. Meanwhile, Apple is preparing a major redesign of iOS, which is expected to debut at WWDC 2025.
While the road has been rough, Apple hasn’t given up. With a complete rebuild underway and lessons learned, Siri might finally be on the path to becoming the smart assistant it was always meant to be.
While the road has been rough, Apple hasn’t given up. With a complete rebuild underway and lessons learned, Siri might finally be on the path to becoming a good smart assistant(or not).
