Skip to content
IFAI

AI tools

Apple tests natural language generation for Siri in tvOS beta

Published Reach 34

Apple just began testing natural language generation features for Siri under the codename "Bobcat" in the tvOS 16.4 beta. The feature is currently limited to telling jokes and handling timers on Apple TV, though code exists across iOS and macOS. It is aimed at upgrading Siri's decade-old template system rather than building a ChatGPT clone.

Apple tests natural language generation for Siri in tvOS beta
Image: 9to5mac.com

While the entire tech world is losing its mind over generative chatbots, Apple is taking its own sweet, cautious time.

According to an exclusive from 9to5Mac, Apple is actively testing new natural language generation capabilities for Siri. Codenamed "Bobcat," the framework has quietly appeared in the latest tvOS 16.4 beta, though its initial implementation is about as low-stakes as it gets.

Starting small on Apple TV

Right now, Apple has only flipped the switch for this new engine on Apple TV. And what is this cutting-edge language tech actually doing? It is helping Siri tell jokes.

Beyond punchlines, Apple is also experimenting with using natural language generation for everyday tasks like timers. Basically, the company is sandbox-testing in areas where a hallucination or glitch won't ruin your day.

Don't let the Apple TV pilot fool you, though. 9to5Mac found that the underlying Bobcat code is already baked into the firmware for iPhone, iPad, Mac, HomePod, and Apple TV. It is just disabled everywhere else. The discovery backs up a recent New York Times report that Apple engineers were testing language-generating concepts, confirming that the groundwork is already in place across Apple's entire ecosystem.

Ditching the decade-old template

Before you get your hopes up, Apple is not building a ChatGPT clone. Bloomberg's Mark Gurman previously noted that Apple is not actively working on a standalone chatbot, and 9to5Mac confirms the current priority is simply rehabilitating Siri.

For more than a decade, Siri has been chained to a rigid, template-based database. When you ask a question, engineers have to have anticipated your exact phrasing and manually built a response slot. That brittle architecture created a massive scaling bottleneck and turned Siri into the internet's favorite digital punchline.

By layering in natural language generation, Apple wants to replace static templates with dynamic responses, letting Siri sound more natural without fundamentally changing what it does. There is no official timeline for when these upgrades will reach your iPhone or Mac.

Why it matters

For everyday Apple users, this is welcome news, but temper your expectations. Testing joke delivery on a television box shows that Apple knows its current models are nowhere near ready for prime time. Do not expect Siri to suddenly write your emails or solve complex logic puzzles tomorrow; for now, there is nothing actionable for users to try.

From a product strategy perspective, this approach is classic Apple. While rivals are racing to ship flashy chatbots and inevitably dealing with factual blunders, Apple is quietly using generative tech to pay down a decade of technical debt. Burying the code across all operating systems while managing activation from the server gives Apple a safe, invisible testbed to gather data without risking front-page PR disasters.

However, the strategy carries real risk. The rest of the industry is fundamentally reinventing operating systems, search, and personal productivity. If Apple spends the next two years just figuring out how to make Siri set a kitchen timer without sounding like a robot, it risks winning yesterday's battle while losing tomorrow's platform shift.

Sources (1)

  1. [1]
    9to5mac.com(opens in a new tab)Original9to5mac.com

Best of AI, daily