Fake AI 

Edited by Frederike Kaltheuner

Meatspace Press (2021)

Book release: 14/12/2021

This book is an intervention - 

Chapter 7

How (not) to blog about an intelligent toothbrush

By James Vincent

“I’d found this dilemma to be a common problem writing about AI. The term “AI” itself is so rotten with ambiguity, so overburdened with varying connotations and expectations, that it cannot be trusted to support much linguistic weight.”

News of the world’s first toothbrush with artificial intelligence arrived in my email in January 2017. It was part of the tide of announcements accompanying the start of the world’s biggest tech trade show, CES; an annual inundation that floods journalists’ inboxes with the regularity of the Nile, leaving behind a rich and fertile substrate that’s perfect for raising some blog posts.

It was not the stupidest pitch I received that year, nor was it the first time someone had tried to sell me a gadget that claimed to have “embedded AI” when it very obviously didn’t. But it sticks in my memory all the same. There was something so nonsensical, so inadvertently Dada about the phrase, that I remember getting up from my desk and making a cup of coffee to process it in mild wonder. The world’s first toothbrush with artificial intelligence, I reflected, waiting for the kettle to boil. The world’s first toothbrush with artificial intelligence, I pondered, returning the milk to the fridge.

Now, I freely confess to occasionally writing tech stories with ridiculous headlines just for the pleasure of lobbing said headline into the world like a cantaloupe into the Grand Canyon. Any tech journalist who says they don’t do this is either lying or has forgotten how to have fun. People love to dunk on stupid gadgets and tech writers love to help them. (Remember: it’s not clickbait if you can get the gist of the story from the headline alone. If you read, groan, then click to find out more about this or that stupid gadget out of a sense of morbid curiosity, then that’s just a good headline.)

To write these headlines, most of the time you don’t need to do more than state what it is the gadget does. The whole point of this sort of coverage—indeed, one suspects, the whole point of this sort of gadget—is the self-evident silliness. The products become more enticing the more superfluous they are, like a variant of a Veblen good in which demand for an item increases as it becomes more expensive. From a journalist’s point of view, this usually means you can avoid adding snark to the headline and let the thing’s idiocy speak for itself. People on social media will quote-tweet you with delight, pointing out the daftness of the product and perhaps adding some comment about “late capitalism” for good measure. All this as if the Victorians didn’t also fill their lives with air-conditioned top hats and mechanical hairbrushes. “Kohler’s smart toilet promises a ‘fully-immersive experience’” is one such headline I remember with fondness from CES 2019—the silliness of the semi-submerged pun matching, I hoped, the product’s stupidity.

It’s still important, of course, for tech journalists to write responsibly about even idiotic gadgets, and two things should be kept in mind when creating this sort of coverage. First, whether the product in question is so obviously a scam or so potentially harmful that covering it in any way is unethical. And second, the degree of credulity with which you write about claims being made about the technology. Is this a silly gadget or is it an actual con?

For the AI toothbrush this left me with a bit of a problem. (I’d decided while drinking my coffee that it was too foolish not to write up.) It seemed like a real product, no worries on that score, but I was uneasy about repeating claims about the device’s “embedded AI”. Would people get that this was a stupid thing to say, or would they think, given the level of ambient hype surrounding artificial intelligence, that this was a legitimate breakthrough of some sort?

I’d found this dilemma to be a common problem writing about AI. The term “AI” itself is so rotten with ambiguity, so overburdened with varying connotations and expectations, that it cannot be trusted to support much linguistic weight. Depend on your readers to parse the phrase “artificial intelligence” this way or that and its meaning will collapse underfoot. This is both symptom and cause of the febrile atmosphere that surrounds discussion of AI—a misunderstanding that is productive for generating hype and selling snake oil, but not much else. If an artificial intelligence researcher, tech journalist, and lay reader all have different expectations when reading those two letters AI, clear communication about the subject is always going to be difficult.

So what to do about the toothbrush? As I read through the press release, Googled the company that made the gadget and scoured their website for something approaching detailed tech specs, it seemed—as expected—that what was being passed off as “artificial intelligence” was just a teaspoon of embedded computation: the thing was only smart compared to rocks.

What it did was use sensors to record when it was used, how long for, and, roughly, the parts of the mouth covered by each brushing. It refined this data into graphs and charts and sent the resulting insights to a smartphone app. It then delivered revelations along the lines of “brush more” or “brush less” depending on how much you brushed. It was the sort of information that a stopwatch could provide without making any claims to embedded intelligence. A little further Googling even revealed that the gadget’s creators had announced a near-identical product in 2014. The only major change was that they’d dropped the term “smart” in favour of the more fashionable “AI”. It seemed this was just another company climbing aboard the artificial intelligence hype train, but what else had I expected?

In the end, I wrote a short little blog post, barely 300 words, with the headline “This smart toothbrush claims to have its very own ‘embedded AI’”. I hoped this formulation—the scare quotes and “claims”—would neutralize any potential deception, but I’m not sure it really did. Later in the day, a little annoyed I’d covered the thing at all, I wrote a longer article castigating superfluous “AI gadgets” in general. More searching dug up “AI-enhanced” alarm clocks, TVs, beds, headphones, washing machines, dishwashers, fridges, and more—each product trading off the same ambiguity surrounding artificial intelligence to sell basic digital functionality as something exciting and new. Plenty of these gadgets attracted press coverage—some critical, some credulous—but browsing through it all I felt that no matter what had been written, it only added to the confusion. Taken together it all seemed like epistemological chaff—meaning thrown to the winds. The world’s first toothbrush with artificial intelligence had only ever been a single point in a scattered and chaotic melée.

James Vincent is a senior reporter for The Verge who covers artificial intelligence and other things. He lives in London and loves to blog.

Next: Chapter 8
Learn to take on the ceiling

by Alexander Reben

Instagram        Twitter