‘AI’ Has A Serious Bullshit Problem

I love a game of “buzzword bingo” as much as the next person but I’m getting a little sick of the way companies are using the terms “AI” and machine learning to describe, well, basically everything. In many cases, the terms are being completely misappropriated. And it turns out I’m not the only one that’s noticed. A new study has revealed a whopping 40% of European companies that call themselves “AI startups” are talking rubbish.

Forbes’ Parmy Olson found the data buried deep within a report from London-based venture capital firm MMC. The report presented the data by saying: “In approximately 60% of the cases – 1580 companies – there was evidence of AI material to a company’s value proposition”.

But the other 40% are misrepresenting things a little. And it’s not just startups either. Earlier this week, we reported on how the mobile version of Microsoft Excel can now import data from an image into a spreadsheet. Microsoft called this AI. I noted that OCR has been around for a long time and calling this AI was pretty disingenuous.

One of the challenges is that AI is nebulously defined. I noted a while ago that there are varying levels of AI and that simply using an algorithm or some degree of automation doesn’t constitute AI or machine learning.

The important lesson here is to look beyond the claims of companies and delve a little more deeply into what they’re really doing and selling. To the uninitiated something like OCR can look intelligent. But it’s important to critically challenge what we see and read and not just accept what vendors tell us.

The full report from MWC can be read here.