Betting on AI will take Scotland backwards not forwards
Going all-in on AI, which is the current strategy in Scotland, will take the country backwards not forwards. It will result in a Scotland that’s less independent, less green, and left with shoddy and second-class public services. Avoiding this is likely to depend on resistance from ordinary workers and communities.
The current AI strategy for Scotland makes a lot of the same fundamental mistakes as the UK government’s own strategy. It positions AI as the most important engine for innovation and growth in the present moment, and therefore key to Scotland’s future. This is buying into the hype, and betting that AI can solve all sorts of tricky structural problems that the Scottish government, like others, is short of answers for. Unfortunately, the kinds of ‘growth’ that AI enables turns out to have toxic consequences.
It’s true that AI’s promoters in Scotland aren’t completely blind to the risks, so they’re committing to an AI that’s trustworthy and ethical. This would be fair enough if it had any chance of happening, but it doesn’t. For starters, the technology behind current AI is fundamentally untrustworthy. AI is based on large-scale pattern finding, and on making probabilistic predictions based on what it finds in the training data. Because of the giant scale of data and computing involved these guesses can be pretty good, but that’s all they are.
By now, pretty much everyone has played with tools like ChatGPT enough to know that AI can be plausible in a way that’s genuinely impressive, but it can also produce utter garbage. This isn’t a fault that can be fixed because, to the machine, they’re all the same kind of speculation. The same applies to all contemporary AI, whether it’s generating funny images or analysing cancer scans.
Building vital services on sketchy tech already seems ethically questionable, but AI falls at the ethical hurdle in other ways. When it comes to anything important, everyone wants AI to be fair and not biased. The problem is, AI is maths not judgement. It turns out there are many mathematical ways to define ‘fairness’, and they all contradict each other. Justice simply isn’t something you can automate into a machine.
Any hope of a responsible AI is blown out of the water by the other ethical disasters that are built into it, including the dependency on taking data without asking, on exploited labour in the global south to massage all the data, and the control of AI by Silicon Valley billionaires with very dubious visions of the future.
For Scotland, one of the biggest of these downsides is AI’s giant environmental footprint. AI depends on vast numbers of computers churning away in air-conditioned warehouses called data centres, and current developments in so-called generative AI, like chatbots, have sent the demand for data centres spiralling upwards.
Each one of AI’s hyperscale data centres demands electricity to power the computers and water to cool them, and that means less water and energy for other purposes. Cities in the American Midwest of the USA are struggling for water because data centres are sucking up so much, while new house building in parts of London was put on hold because data centres had already seized so much of the electricity grid.
At the end of the day, AI is yet another extractive industry. It depends on extracting data and labour, and on the very physical extraction of energy and water, not to mention conflict minerals for the computer chips. Betting on AI for Scotland is like going back to the fossil fuel industry, and trashes any chance of a just transition that’s fair to workers and fair to the planet.
AI isn’t the answer to the climate crisis, it’s part of the problem; even when it pretends to be ‘green’, as in Ireland, it’s because it’s grabbing all the renewable energy that could otherwise go to local communities. And we’ve already seen what this means for jobs; at the first opportunity, the instinct of corporations and governments is to think how AI can help them replace workers without so much as a thank you and goodbye, even if the result causes more problems in the longer term.
Just as bad are the claims that AI’s fakery could fix all sorts of real problems for the people of Scotland. Waiting lists for healthcare? We’ve got bots that can fix that. Not enough teachers, no problem; every child will get their own personalised AI-genius tutor.
This is all a giant exercise in avoiding the tricky issues of resources and the restructuring that it would take to share wealth more fairly. All the evidence is that AI-powered services for health, education or anything else will be shoddy and second-best, and you can be sure they won’t be the first choice for people with money to pay for better. Even stock market investors are starting to panic about whether AI can deliver on the billions they’ve invested, and a danger for Scotland is that the industry will milk the public sector to prop up its failing claims.
Instead of swallowing blather about Scotland as an AI powerhouse, we should be asking who really benefits, and who really controls all of this. AI is built on resources that Scotland can never match; Microsoft alone spent as much on data centres in the last quarter ($19 billion) as Scotland’s health and social care budget for the entire year. AI software and hardware is totally controlled by rapacious Silicon Valley corporations led by figures who largely believe democracy has had its day and it’s time of the tech overlords to take over.
Going all in on AI is a surefire way to make Scotland less independent and less democratic. Whatever the social issue, AI is almost certainly the wrong answer, and it’s time for workers and communities in Scotland to step up and ask the difficult questions.
A just transition for Scotland, and a greener and more independent country, is a future that depends in part on finding alternatives to AI. Luckily, there are strong enough movements for social and environmental justice in Scotland, and strong enough histories of fighting for that justice, to give this alternative future a chance.
Back in ancient Greek times, a group of intellectuals would sit around under the stars well into the night and solve things between them. People like Mozart, Beethoven, and Wagner could look at a music score and hear the music in their heads. Thirty years ago I had in my head, 20 or more telephone numbers. We don’t need any of that stuff now, it’s all done for us. It’s a creeping problem and we are the frog in the pot of lukewarm water over the burning sticks.
Computers are getting smarter, they may or may not be intelligent because we don’t know for sure what intelligence actually is or where it comes from. One thing is for sure, computers will continue to get smarter and when quantum computers get better and their smartness starts to grow exponentially, then we’ll have problems. And if the genie gets out of the bottle then who knows where we’ll end up???
Aye to that Dan and yon Tony Blair Institute for Global Change… with his son Euan a tech mogul too… not only privatising our healthcare but also going big on the bots doing the job !
Thanks for calling this out, especially in terms of ecological footprint.
Fortunately the energy and resource requirements of AI are so vast that it will never deliver on the scale that those Silicon valley investors have fantasies about. Some years ago it was predicted that the internet alone would use up a quarter of the world’s electricity by 2025 – which you can do – but how many hospitals do you shut down to allow that to happen? How much street lighting has to be turned off to allow that to happen? etc. And AI would be in addition to that.
It is likely that the vast amounts of AI deployed will revert to the entertainment industry, fake pop stars or re-animated dead ones, computer art, porn, 3d cats and adverts projecting out of your smartphone, n’ all that gubbins.
The simplest solution, as Sam Mitchell over at Collapse Chronicles occasionally reminds us, is to simply not take part and not buy any of that corporate stuff. If you don’t buy and use a smart phone, nor have the internet, your interaction with AI will be pretty much zero. And more importantly, your contribution to corporate profits will be zero.
At the end of the day, all this technosphere requires a gullible public buying into the advertising on sufficient scale for the corporations to consider it worthwhile. Without us, they are nothing. All we have to do is persuade the rest of the public not to participate……. (gulp!)
Most of this may be true (there are conflicting views on how to categorise Artificial Intelligence versus Machine Learning), and the prime applications may be to play stockmarkets, rig politics, run surveillance and assassination programmes, and modify human behaviours for vile, pecuniary and/or petty purposes… but I’d still put the research in, for those very reasons, to counter such effects, as long as you make the research public.
The prospect of general AI is sadly only likely to usher in an era of virtual psychopaths even worse (for the biologically living planet) that the ones making them, in the short term (and that’s maybe all we’ll have, especially if AI is militarised).
I do see a positive role for AI in government, but it has to be a white box one, as far as possible.
https://en.wikipedia.org/wiki/White_box_(software_engineering)
For example, uses in climate and pollution modelling. And further down the line, giving voice to the interests of non-humans (and future generations).
#biocracynow