Dear Google, don’t pull an Apple and advertise features you can’t deliver

At Google’s developer conference this week, AI was everywhere and being shoved into everything. Generally analysts were happy with the company’s performance and promises, but what’s ultimately more important is whether Google will actually be able to execute all that AI supposedly “coming soon.”

For a cautionary tale, look no further than Apple, which at its developer conference last year made many similar promises. Then, like now, AI was going to be integrated across apps and its smart assistant Siri was going to become truly smart. The iPhone maker promised the holy grail of UI: having your device correctly anticipate and act on your needs and wants.

The goal was to “make Siri more natural, more contextually relevant, and, of course, personal to you,” Apple AI and machine learning project manager Kelsey Peterson said at the time. The assistant, recreated from the ground up with AI, would be able to pull from your texts and emails and cross-reference real-time flight tracking, for example, to tell her when to pick up her mom from the airport.

The promises went on:

“Thanks to Apple Intelligence, [Siri] has awareness of your personal context. With its semantic index of things like photos, calendar events and files, plus information that’s stashed in passing messages and emails like hotel bookings, PDFs of concert tickets and links that your friends have shared, Siri will find and understand things it never could before. And with the powerful privacy protections of Apple Intelligence, Siri will use this information to help you get things done without compromising your privacy. You’ll be able to ask Siri to find something when you can’t remember if it was in an email, a text, or a shared note, like some book recommendations that a friend sent you a while back or for times when you’re filling out a form and need to input your driver’s license, Siri will be able to find a photo of your license, extract your ID number and type it into the form for you.”

One year later, ahead of this year’s Apple developer conference, much of what was advertised hasn’t panned out. Many of the promised features aren’t available and those that are don’t work as promised. Perhaps as a result, Apple Intelligence failed to drive a meaningful iPhone upgrade cycle, Apple is behind its peers in the AI space, and Apple’s head of services, Eddie Cue, has warned, “You may not need an iPhone 10 years from now.”

When the company’s software chief, Craig Federighi, tested the upgraded Siri weeks before its planned release in April, “he was shocked to find that many of the features Apple had been touting — including pulling up a driver’s license number with a voice search — didn’t actually work,” Bloomberg reported recently. The promised Siri upgrades are still “months away from shipping,” and Apple doesn’t plan on discussing them much at this year’s WWDC. Indeed, the company will stop announcing new features more than a few months before their launch.

Let’s look now at Google’s recent promises. Here’s Josh Woodward, VP of Google Labs and Gemini, at the developer conference Tuesday:

“Our goal is to make Gemini the most personal, proactive, and powerful AI assistant, and it starts with being personal. What if your AI assistant was truly yours... an assistant that learns you, your preferences, your projects, your world, and you were always in the driver’s seat?”

He went on, demonstrating a number of features that would certainly be nice to have, but felt a bit like a bridge for sale.

“See today, most AI is reactive: you ask, it answers. But what if it could see what’s coming and help you prepare even before you ask? Imagine you’re a student you’ve got a big physics exam looming and instead of scrambling, Gemini sees it on your calendar a week out, but it doesn’t just remind you. It comes with personalized quizzes crafted from your materials, notes from your professor, even photos, handwritten notes. That’s not just helpful. It’s gonna feel like magic.”

Yes, of course — if it works. As we learned from Apple, getting this stuff to actually work is a far cry from conjecturing in a demo. Truly useful AI, it turns out, is hard. Apple’s Genmoji, where you use real language to make custom AI emoji, take a long time to generate, usually don’t look that great, and can even overheat your phone. Apple’s text summaries of news were shut down after not just being comical but flat-out wrong. Siri’s upgrades haven’t materialized and the assistant can seem worse than ever.

We’ll check back in this summer when “personal context” is slated to roll out across Google products to see if it’s all it was talked up to be.