Pixel Envy

Written by Nick Heer.

My WWDC Wish: Siri That Does Not Suck Ass

Becky Hansmeyer:

There are things to celebrate even in hard times, and WWDC is when we all get to celebrate Apple engineers’ hard work over the past year. I really enjoy reading wishlists and predictions, so this year I’ve compiled a WWDC 2021 Community Wishlist. You’re welcome to contribute, just submit a pull request (or send me a note on Twitter and I’ll add it for you).

Michael Tsai compiled a bunch.

I know it is the Wednesday before WWDC and, so, a wish list does not make sense. There is simply no time. So let’s call this a single wish I hope is granted: I want to be able to depend on Siri for more than timers and reminders.

While driving out of Calgary on the way to Lake Louise today, I found myself a little disorientated after being rerouted by a series of construction zones. So I asked Siri to “get me directions to Lake Louise”. Rather than showing me how to get to one of the most famous lakes in the world, or its namesake town, or anything else in the vicinity that shares the lake’s name, Siri gave me directions to Lake Louise, Alaska.

I tried again: “hey Siri, get me directions to Lake Louise, Alberta”. It responded with a route to the hamlet of Parkland, Alberta, nowhere near Lake Louise and in the opposite direction. I gave up.

This is not an exceptional case for me or anyone I know who uses Siri for more than truly ground-floor things — timers, reminders, dictating text messages, and the like. This is the norm for Apple’s supposedly intelligent assistant.

Being unable to trust Siri, especially behind the wheel, is distracting and can be downright dangerous. I am a reasonably cautious driver; I have never touched my phone behind the wheel. Features like CarPlay and Siri are supposed to help drivers safely use an iPhone for driving-related tasks. This is not funny any more. Siri is ten years old in Apple’s iteration, and even older as independently developed. Apple has clearly put a lot of work into its voices and speech recognition, but it is a deeply disappointing product all the same. I would rather have a stilted robotic voice responding accurately than a high-quality voice that cannot seem to do the obvious and correct thing.