AI TRİSM

New computing paradigms come with new challenges. Steve Cassidy explores how to use AI safely and securely

- 

I’ve heard of a prism… what on earth is a TRiSM?

Your uncertainty is understandable, as it’s a rather forced acronym. Simply put, it refers to the particular human and organisational issues raised by AI adoption projects, namely Trust, Risk, Security and... well, the “M” just stands for Management. Which is fine, because the first three points are quite big enough to require attention and consideration on their own.

I can see why trust is an issue – ChatGPT talks a lot of nonsense.

Businesses wanting to embrace AI certainly need to think about how to deal with unreliable output – and indeed how to detect it. Old tales from the 1970s come to mind of bosses ordering employees to double-check computer calculations by hand.

But the problem runs deeper. The inner workings of AI algorithms are by design completely hidden, so underlying problems may be almost impossible to diagnose or fix.

What about risk – is that the threat of an AI taking over the world?