Thread by @edwinhayward

I'm catching up on the last question of #bbcqt on artificial intelligence, and it's a reminder that there's huge danger in having panels discuss topics that no panellist really understands anything about.

(And there's that perennial mistake of mixing up AI and robots too. Gah!)

Add to that the risible Tory boosterism at the idea that a drop-in-the-ocean £100 million investment can make the UK one of the global leaders in the AI sector.

Desperate, desperate stuff from start to finish.

The audience, so far, are just as lost.

It's like watching a busload of people who can't swim get dumped in the middle of the Atlantic Ocean to fend for themselves.

As for the BBC producers, it's depressing that they drew on such old examples.

The output of generative AI today is better than a week ago, better again than a fortnight ago, and much much better than a month ago.

AI years are like dog years, multiplied to the nth degree.

All in all, the BBC would have performed a public service by blanking out that 15 minute segment entirely and replacing it with the old test card (remember those?)

There was no intellectual value to be wrung from it at all, and it will have left the average viewer less informed.

I asked Bing AI a reformulated version of the question. The answer it gave me was enormously more useful.

(It's up to you whether you feel it was cheating for me to rephrase the original question. I was merely trying to contextualise it and ensure an up-to-date response.)