If you’ve been watching TV of late, you’ll likely have seen the Microsoft AI advert where an actor makes ambitious claims about “ay-aah”.

Every time it comes on I muse about how a company known for ’undocumented features’ could expect me to believe it will be able to change the world with its super technology. Then I remember the Skynet computer system from Terminator that destroys the world and realise that not all change is good.

Irritating adverts aside, I don’t doubt that AI is going to transform everyday life. But there are pitfalls waiting behind what appears to be a seductively powerful tool. I was invited to a conference few weeks ago and AI was the main topic of discussion. Ken Hughes’ talk on the future impact of the technology left quite an impression on me but raised more questions than answers.

“I doubt there’s going to be much enthusiasm towards talking to jumped-up Speak & Spell machines”

Sales of voice control hubs and the internet of things demonstrates there’s an appetite for AI in our homes, even if Alexa is apt to order fork handles instead of four candles. In an interesting twist, we’re even seeing these devices helping people with dementia navigate their everyday lives.

But anyone watching that YouTube clip of a Google Assistant booking a hair appointment, without the salon staff being aware that they were talking to a machine, can’t fail to be both impressed and a little disturbed by how deceptive this technology has become.

This raises ethical issues. In the case of the hair salon, should the staff have been told beforehand that they were talking to a computer? Perhaps more importantly for all retailers, should customers being served by a non-human entity be aware that the personal service they’re receiving isn’t as personal as they think it is? What happens when my digital assistant talks to yours and they make lightspeed decisions about our lives without our knowledge or approval?

We’re already seeing AI sitting behind online help bots with customers believing they’re talking to Gary in sunny Scunthorpe, when in fact they’re chatting to HAL9000 somewhere in cyberspace. Open the iPod bay doors please HAL!

When AI goes off-script

With the backlash against non-UK call centres literally ringing in our ears, I doubt there’s going to be much enthusiasm towards talking to jumped-up Speak & Spell machines. I’m frequently irritated to the point of meltdown just by automated call-handling systems.

There have already been high-profile cases of AI platforms going spectacularly off-script leading to PR disasters. Three years ago, Microsoft had to shut down an experimental AI after it went on a racist rant online. Imagine the impact that sort of thing could have on a high-profile retailer.

AI algorithms are still pegged to basic human assumptions. We’re a long way from being able to rely on this technology not to do just as much damage to our businesses as any other rogue employee after a lunchtime bender. Even more so, given the speed they operate at.

“Personal customer interaction is increasingly being seen as the antidote to a remote online experience”

That’s not to say that we shouldn’t embrace AI, just that we shouldn’t be relying on it to make decisions we’re not ultimately in control of. The software problems that have been identified as the cause of some recent tragic air crashes demonstrate what happens when machines think they know what’s best for us better than we do.

Moreover, in a world where personal customer interaction is increasingly being seen as the antidote to a remote online experience, consumers might become a tad paranoid if they discover they’re often interacting with a decision tree of preformatted responses instead of a real person. Consumers want the human touch, not a touchscreen. Artificiality is not a great service USP.

To slightly misquote that other well-known Hollywood embodiment of AI, customers faced with an insincere and impersonal interactions with our businesses could soon be telling us: “I won’t be back.”