Hi ,
Are you familiar with the concept of "AI-washing"?
Techopedia defines
"AI-washing" as "a deceptive promotional practice that exaggerates or outright lies about a product or service’s use of AI". You've undoubtedly come across it before - whether you're aware of it or not.
The idea is that companies or brands associated with AI are perceived to be more valuable or advanced, motivating some companies to mislead customers into believing their products
incorporate AI.
Much has been written on this topic recently, in the wake of the recent revelation that Amazon was using human workers to "power" the AI technology behind their Just Walk Out checkout-free supermarkets.
Yet, the irony is that there are many
situations where people feel more comfortable knowing that humans are overseeing AI.
Last week I asked readers about their views on this topic and most readers were either uncomfortable with the use of AI without human oversight, or only comfortable in low risk situations.
Some of the comments I received include:
"If it's a digital creation thing, I'd want to approve it. If it's a physical action thing, I'd want to be able to override it. If it's a suggestion thing, I'd want to be able to accept/reject it." - James Turner
"I would think that it would depend on the type of task in terms of a risk category. If it is a high risk AI Task - let's say for example that could affect a human's health then I think the human intervention threshold should be low." - Ricardo
Jurawan
"If the trust or confidence is not there, then no go. Applies to AI or humans." - Darren Wu
Here's the
thing...
There are circumstances where most people would be very concerned to discover a human was doing the work of AI.
While you may be fine with Grammarly's AI reviewing everything you write, who wouldn't be horrified to learn a human was doing the
same?
And there are also circumstances where you would be equally concerned to discover an AI tool was doing the work of a human.
It all comes down to context.
There is nothing wrong with starting with a manual process and progressively introducing AI to reduce the burden on human workers. Producing fully AI-powered products that can function with an acceptable level of accuracy, under all circumstances, is hard - and people get that.
But nobody likes to be fooled.
The problem with AI-washing isn't the use of humans in place of AI. It's not being honest when you do.
Talk again soon,
Dr Genevieve Hayes.
