Apple is trying to fix what it overhyped

APPLE

Apple’s first wave of AI tools didn’t quite land. Now, a new Bloomberg report lays out where it all went wrong, and how the company’s now trying to fix it.

A big part of that fix is rebuilding Siri from the ground up.

The new version, known internally as LLM Siri, is being completely reworked using a large language model.

Apple’s original plan to just stick generative AI onto the old Siri system turned out to be more trouble than it was worth.

So what went wrong?

  • Senior leadership hesitated to go all in on AI early, especially when it came to buying the expensive GPUs everyone else was racing for.

  • Apple started late. The AI push only kicked off after ChatGPT made waves in late 2022.

  • The company’s AI marketing overpromised before the tech was ready, leading to delays and disappointment.

Not Siri 2.0… Siri reborn

Now, Apple’s Zurich-based team is building a new system from scratch, aiming to make Siri actually feel like a modern AI assistant, more natural in conversation and better at pulling useful information together.

They’re also using on-device data comparison (via differential privacy) to train AI in a more private way, and they’re exploring turning Siri into a web search assistant, possibly through a partner like Perplexity.

Meanwhile, Apple’s AI head John Giannandrea has stepped back from Siri and product development.

He’s staying on at the company for now, but it’s clear Siri’s future will be shaped without him.

Zurich team, better be coding like their lives depend on it.