Using AI in business? Proceed with caution

Embracing AI is tempting, but where’s the safety net?

That quote from The Imitation Game, the film dramatising the quest by mathematics genius Alan Turing to build a machine to break an ‘unbreakable’ Nazi code in WWII, captured my imagination.

Because it is true isn’t it. Sometimes we come across things that seem unpromising or innocuous only later, with the benefit of hindsight to realise how significant they were to become.

Not only did Turing break the Nazi Enigma code, saving millions of lives and ending the War an estimated three years earlier than might otherwise have been the case, but also his ‘Turing Machines’ became the problem-solving technology that today we call… ‘computers.’

Just as Turing’s ability surpassed the expectations of those around him, so did the technology he gave birth to. And the implications, though unintended, were greater than perhaps he ever could have imagined.

Similarly my own week started innocently enough…I flew to Newcastle on Monday night and spent Wednesday and Thursday at a two-day leadership event. Good discussions, plenty of insights, and lots of learning but nothing out of the ordinary.

Back home, the weekend routine took over. Saturday saw me cheering on the kids at athletics. Today, though, is special: Lucy and I are celebrating our wedding anniversary. Thirteen years married (21 together) and still going strong.

However, then something extraordinary caught my eye. It was Sky News experimenting with AI…

A presenter had asked AI to summarise content from his podcast. Simple enough, you’d think. But what came back wasn’t even from his podcast.

The presenter tried to correct it, pointing out the error. The AI argued back, convinced it was right… even claiming the content was submitted at 8:30am, although it was still only 7:50am! A surreal debate with a machine.

Funny, perhaps, but it raises serious questions…

Businesses everywhere are keen to integrate AI. The potential is obvious, and the scope is huge. But with greater reliance comes greater risk. We already know AI makes things up, so how comfortable are we basing business decisions on it?

There’s also the bigger concern around data privacy.

What happens when sensitive commercial or personal information gets entered into AI? Do we know exactly where it’s stored, or how it’s used? Some AI platforms have allegedly been selling user data to third parties. If that’s true, how safe is our information?

From a GDPR perspective, we need answers before diving head-first into tools like ChatGPT or Microsoft Copilot. Is it lawful and fair to share the data we’re entering? Have the individuals whose data we input given consent? Are we transparent enough about how their information is being used?

We also need to know how long this data is retained, who accesses it, and for what purposes. Most importantly, can we demonstrate accountability if something goes wrong? Machine learning evolves constantly so what stops AI from regurgitating sensitive information to someone else who shouldn’t see it?

And above all of this, I have this niggling question in the back of my mind… How smart is it to empower machines that argue back, will not admit error without significant pressure and have no compassion.

I’m all for innovation, but these questions matter.

Until there’s clarity on checks, balances, and accountability, businesses need to proceed carefully. AI offers plenty of benefits, but only if we can trust it. And as my story illustrates, right now we KNOW we can’t.

Peter Fisher

Director of Risk & Intelligence