Facing the music: our love affair with AI is about to be severely tested
First things first. I don’t have any doubt that AI is set to transform our lives. In fact, it’s clearly one of the dominant issues of the coming decade, ranking alongside climate change and global conflict in terms of its impact.
But climate change and global conflict aren’t exactly positives, are they?
In fact, they’re the kind of things that tend to keep a lot of us awake at night.
I want to believe that AI is in a different category, but nothing I read or view tends to fill me with great confidence.
So when people talk about the huge benefits of AI, what do they really mean? Or, to look at it another way, in ten years’ time, when we’re giving it the Monty Python test — ‘What has AI ever done for us?’ — will we be able to point to the aqueducts, sanitation and great wines?
Medicine and science, to my mind, are probably the areas where there’s the most plausible case that the technology has potential to do actual good. Spotting patterns that are unobservable to humans amid vast quantities of data and developing novel solutions.
At the end of last year, for example, it was reported that bots had found a cure for a drug-resistant bug. So hats off to our algorithmic pals for stuff like that.
Even so, when we realise that an AI is better than diagnosing us than a doctor, where does that leave our relationship with medics? And when the practitioners discover that AI has a better bedside manner too, they must be experiencing a sense of existential dread. The technology is already challenging fundamental assumptions about the value of human beings and what we are able to offer each other.
Clearly there are numerous applications in business, which we could see as potential ways of driving growth and stimulating economies. There seems to be a lot of excitement, for instance, about the generation of ‘synthetic data’, which has a number of possible uses. In my own field of advertising and marketing, it can allow you to test the effectiveness of, say, potential strategies and media choices, no doubt saving agencies and their clients piles of cash.
According to Forbes, AI will reduce human error, automate repetitive tasks, improve processes and facilitate decision-making. All the kind of stuff that sounds like motherhood and apple pie to corporate executives, but realistically spells the end of a lot of human jobs.
So forgive me for being somewhat cynical about all this.
I suspect the main beneficiaries will be the large corporations in the best position to scale the technology right now. As a result, I can see a few very big companies getting even bigger and a few very rich individuals getting even richer, while governments play an endless game of catch-up.
For the likes of you and me, I reckon the AI revolution may not look quite so rosy. Yes, I know this runs contrary to the spirit of debate on platforms such as LinkedIn. But the whole culture of these business networking sites is always to embrace change as something inherently good and to dismiss as Luddite or difficult anyone who sticks their hand up in the corner.
If time travel technology were invented, there would be a frenzied discussion in the media about the potential dangers of changing the course of history, obliterating our own grandparents and destroying the timeline that we have come to know. On LinkedIn, people would be telling us to focus on the positives. Time tourism could prove to be a big growth market and if we screwed up our next TED Talk, we could go back and do it again.
While I don’t think the existential threats from AI can be completely dismissed, they’re not actually my main concern. The technology is producing too many alarming phenomena right now for us to waste time worrying about the end of the world. And these immediate issues are almost all the product of so-called ‘generative’ AI, which increasingly is in the hands of the whole world and his demented dog.
With just a small sample of someone’s voice, you can create an entirely plausible representation of them. Good enough to fool someone that their loved one is being held for ransom. Good enough for a teacher to falsely paint their boss as a racist.
School students create deepfake porn videos of their classmates, fuelling an ever-growing technological dystopia for troubled teenagers.
In a bid to compensate for generative AI’s tendency to reflect the biases of its training data (presenting business leaders and politicians as white, middle-aged men, for instance), so-called ‘woke’ AI overcompensates by presenting people of colour as Nazis. As our future gets reshaped, history gets reimagined — projecting crimes of the past on to people who had no part in them.
Meanwhile, a British think tank predicts that up to eight million jobs could disappear in the UK in a ‘worst-case’ scenario. Everyone from administrators and customer service representatives through to accountants and graphic designers are under threat. (Weirdly, the IPPR also postulates a more positive scenario in which no jobs are lost, which seems a tad optimistic.)
If we imagine the truth lies somewhere in between, there is a very heavily-laden apple cart that is about to be upset. Because we could be looking at a lot of folk whose raison d’etre is taken away. And that’s before you even consider the fact that people without jobs don’t pay tax and will claim state benefits, so a public policy nightmare looms.
A glimpse of the near future can be seen in the new AI music platforms — Udio (which McAfee decides to block) and Suno. From basic text prompts, the technology produces very plausible musical tracks. The critical thing to note is that they don’t sound any more as if they’re AI-generated and they can actually have quite a listenable quality about them. Compare this to the same tech just a year or so ago and the step change is staggering.
The next stage in this process is not hard to imagine.
As we find we can generate longer pieces of music algorithmically, which are designed to reflect our existing tastes, what becomes of the musical artist? What if I enjoy the AI-generated music much more than anything I happen to hear serendipitously, because it is tailored specifically to me?
Think of the monkeys writing Shakespeare on typewriters. Except now they’re turbo-boosted AI simians, able to produce Romeo and Juliet before you can say carriage return.
These are no longer questions that can be left to science fiction writers. They’re issues we’ll be confronting in the 2020s.
Remember when we used to imagine that science might eventually give humans a lifetime of leisure? Utopians thought that freed from the world of mindless labour, we might spend our hours exploring music, poetry and art — the very things that help give meaning to our lives and allow for self-expression. In fact, the machines are coming for this stuff first.
