In my travels, I’ve seen derelict towns. The reason they were built has passed into history. A frantic fever swept through an area like an unstoppable storm. It might have been precious metals that excited the original residents. Gold rushes feed the desire to get rich quick. It doesn’t take the greatest minds in the world to figure out why gold fever will always have an appeal. The onrush of people joining the throng keeps going until opportunities have collapsed.
Breakthrough technologies, or their potential, can be just like a gold rush. There’s no doubt that 2025 will be a year of such phenomena. Top of the list is Artificial Intelligence AI[1]. If you want to be a dedicated follower of fashion[2], then AI is the way to go. Thank you, The Kinks. Your lyrics are as apt now as they were in the 1960s.
Predications range from the best thing since sliced bread to the end of humanity. Somewhere along that line is realism. Trouble is that no one really likes realism. It can be somewhat dull.
I’ve always viewed advancing technologies as a two-edged sword. On the one hand there are incredible benefits to be reaped. On the other, costs can be relatively unpredictable and devastating. I say “relatively unpredictable” as there’s always the advantage of knowledge with hindsight. Lots of commentators love to practice that one.
In desperation to gain the economic benefits of AI the current utterances of the UK Government may seem a little unwise[3]. Certainly, there’s nothing wrong with wishing to build a significant domestic capacity in this area of technology. What’s concerning is to always talk of legislation and regulation as a burden. Particularly when such language comes from lawmakers.
The compulsion to free-up opportunity for a western style gold rush like scenario has a downside. That is all too evident in the historic records. Ministers in this new Labour Government remind me of Mr. Gove’s past mantra – we’ve had enough of experts. Rational dialogue gets sidelined.
Even now we have seen generative search engines produce summaries of complex information sources that are riddled with holes. This experience reminds me of past work cleaning up aviation accident databases. Removing all those 2-engined Boeing 747s and airport IDs with one letter transposed. Data by its nature isn’t always correct. The old saying, to err is human, is always applicable.
The concerning aspect of AI output is its believability. If error rates are very low, then we stop questioning results. It gets taken for granted that an answer to a question will be good and true. There we have a potential problem. What next. AI to check AI? Machines to check machines? There lies a deep rabbit hole.
[1] https://www.technologyreview.com/2025/01/08/1109188/whats-next-for-ai-in-2025/
[2] https://youtu.be/stMf0S3xth0
[3] https://www.theguardian.com/technology/2025/jan/11/uk-can-be-ai-sweet-spot-starmers-tech-minister-on-regulation-musk-and-free-speech
John Vincent 