In 1999, David Bowie told Jeremy Paxman the impact of the internet would be both “exhilarating and terrifying”. Paxman scoffed at the internet as merely a tool. With the benefit of hindsight, we know Bowie was all too right.
That’s what it feels like discussing artificial intelligence (AI) in 2025. There’s a lot of hype and noise around it, much like with the internet at the turn of the century, but it is still a feral beast. How it will really impact us is yet to be truly understood.
It’s difficult to fathom the impact the internet has had on society. The easy areas to point to are how we consume news and make purchases. Beyond that, it is fundamental to how we operate everything in our daily lives. We look at our phones first thing in the morning. We log on at work. Most of the applications we use exist in the cloud.
Ask yourself this. How many phone numbers do you know off the top of your head? Better yet, when did you last use a physical phone book? One more, just one, when did you last use a paper map to find directions?
Harbouring this knowledge and conducting these activities was normal just two decades ago. Even those with the most far-fetched science fiction expectations of the internet could not have foreseen the assorted flavours of its impact. The hype was there, people were telling us all this would be big but not many of us realised just how ubiquitous it would become.
These days, it’s just there in the background of our lives. Like boiling a kettle, we don’t think about what is required to make the internet work; we just get annoyed if it doesn’t. That’s the journey AI is on right now.
That role in the background, the acceptance that it works, is important. The late psychiatrist Dr Séamus Mac Suibhne told me in 2018 that while young people of today are more used to using digital tools, it doesn’t mean they are more technically competent.
This need to “just work” applies to every technological leap. When you boil a kettle, you press a button. You probably don’t know how to construct a functional kettle let alone the infrastructure required to send clean water to your tap to fill it or power safely to your house to make it boil.
Why on earth would you know this any more than you remember more than a few phone numbers? Replacing the need for active knowledge, becoming an accepted part of the background infrastructure of our lives is the road AI must wander as the internet did before it.
There are already areas in business where the practical application of AI is commonplace; it is just not commonplace in everyone’s lives.
AI is heavily used in banking as a tool for fraud detection in particular, while it is now unusual for supply chain management systems not to use it to a substantial degree. The recruitment sector has adopted it for processing applications and CVs.
The uses most of us currently come across in our daily lives are far less practical and often harmful. Deepfakes are the obvious example while bot armies across social media spread inanity and misinformation with equal measure.
Using the likes of ChatGPT as a search tool can lead to some wildly inaccurate results. With such large language models (LLMs), the goal is to answer first and be accurate second. That’s a headache for academia, which is getting to grips with papers that are being written using LLMs rather than by their authors doing proper research.
So plenty of hype but, thus far, little real impact in most consumers’ lives. AI hasn’t, at least consciously, changed the way we live all that much in 2025. Much as it was with the internet at the turn of the century.
The hype that preceded the dot-com bubble was enough for expectations of the internet to have been seen as excessive. In truth, it was a case of businesses trying to run before they knew what walking was, let alone how to do it.
That’s the stage AI is at now. Skynet and Terminator analogies abound but the real impact on the workforce has not yet been felt.
We know jobs will be lost and new ones created, and not just in high-tech fields, but anyone who tells you for certain what will happen with AI is either a fantasist or about to be exceedingly wealthy. As with the internet, working out what will change as a result of AI’s explosion is still anyone’s guess.
ARPAnet, essentially the seedling for the internet, was first planned in 1958 and came online in 1969. But the internet would not become well known or heavily used for decades to follow. Likewise, AI has existed for many decades but it is only now becoming a household word.
There will be many failures, particular in the near future. Yet the potential is as exhilarating and terrifying as Bowie thought of the internet. In time, AI will become embedded in our lives to such a degree that it will not consciously intrude on our awareness, in much the same way as the internet now.
