AI begone! T3’sgreatest tech expert is still very much human

One explosion in Dr Frankenstein’s lab later…

ILLUSTRATIONS: STEPHEN KELLY

IAIN SHERE, WOTTON-UNDER-EDGE

Q. How scared should I be of ChatGPT?

A. Very scared, and not scared at all. Both at once. Over the years, Guru has done his fair share of monkeying about with AI. He’s generated nonsensical answers to your questions with it, he’s made weird pictures of RoboCop with it, and he’s totally failed to understand how it works or quite why it makes his electricity bill so high. AI and machine learning still feels like unfathomably cool technology, particularly for someone who rose through the ranks in the 8-bit era of computing.

ChatGPT is a banner-waver for the next generation of machine learning. In itself it’s not scary because it is, for all its brilliance, still remarkably dumb. Actually getting it to do what you want is a case of using very specific language and making do with sub-par and often utterly incorrect results. What’s slightly more scary is the tingling undercarriage it gives to executives looking to save a buck on humans; they’re wrong, obviously, but Google can’t yet tell the difference between words written by flesh or by machine. Be aware that less reputable outlets will churn out endless robot dreck until search can tell the difference.