Trending Articles
Trending Topics
Trending
If AI delivers on its promised disruption, workers will lose their jobs, governments will lose income tax, and the whole system will collapse.
Experts have mooted plenty of ideas through the years to avoid this apparent economic dystopia. Completely rework the tax system. Tax data centres or other AI infrastructure. Create universal basic income so taxes matter less. And the old favourite of taxing bots.
A new twist on the latter comes via Charles Radclyffe, a Welsh software entrepreneur, who argues bots should face a minimum wage. (Charles has also written for us on the difference between R&D and innovation and why meetings in the Metaverse are ethical, productive and simply make sense.)
And Radclyffe knows what he’s talking about: his own company EA automates backend work. “Every time we bill [for a month’s AI work], that is a job from the economy gone and moved into a data centre,” he told the BBC.
He also notes that AI will be able to manage data entry tasks in 20 seconds that would take a human two weeks. No one is going to kick up a fuss when handing two weeks of drudgery to bots – unless that work was paying your mortgage.
Radclyffe is one of many across tech arguing for a minimum wage or some sort of taxation system for robots, saying that could be used by the government to “throttle the adoption of AI and create a more level playing field”.
If AI ever works well enough to cause such disruption, holding it back via an arbitrary payment could well be detrimental to the economy and British businesses, with the work shifting overseas – after all, AI doesn’t have to be “based” on these shores. Look at the automotive industry: if the government banned or discouraged the shift to robots, production would still have moved overseas, only faster.
Also consider the ability of many companies – not Radclyffe’s, to be clear – to (legally) dodge paying taxes. Last year, Starbucks once again didn’t pay any corporation tax, handily posting a loss after paying “royalties” and “licensing fees” to itself. We can’t even get these businesses to pay regular old corporation tax, let alone whatever complex tithing algorithm will be needed to calculate the value of bots.
After all, this isn’t as simple as one bot replacing a specific person. It’s a system reducing human work at scale. Base the tax on layoffs, and they’ll say the redundancies are for business reasons, not bots. Switch to a tax on work completed, and suddenly those bots will become a lot less productive on paper. Tax on money spent buying the bots, and watch that line shift in the balance sheet. And so on. Accountants and tax lawyers always win.
Beyond that, we’d have to tax them an awful lot to make up the equivalent cost to society. The laid-off worker is not only no longer able to prop up the economy with spending, but will need costly social benefits like retraining and unemployment support.
We’ll need to bill British businesses an awful lot for them to balance this equation – and in this instance, they’re the ones paying. There’s a reason Big Tech companies like this solution: they still get paid and their customers cover the cost for societal disruption.
So what’s the answer? There isn’t one.
If AI causes this level of disruption – and we’re still in the land of if when it comes to AI – we’ll need a multitude of solutions.
Some sort of universal basic income or other way to ensure people have the means to survive and continue taking part in the economy. A bot labour tax, as Radclyffe suggests, to smooth the shift away from income tax, otherwise the government will quickly fall short. (Look at the struggle to replace petrol tax when it comes to EVs.)
But none of that answers the big problem that is often ignored: by making Big Tech our source of labour, we hand them all our money (and much more besides) now and in the future. So perhaps we’ll add on a data centre tax to claw back some cash from AI developers, and increase corporate taxes for AI companies, both users and providers.
But we should remember that they’re very good at not paying a fair share.
Indeed, one reason Google, Meta, and the rest have the funds to make such big bets on AI is they aren’t taxed properly. And haven’t been for a long time.
A couple weeks back, a report from the Institute on Taxation and Economic Policy (ITEP) revealed that 88 of the largest corporations in the US paid no federal income tax last year, despite earning $105 billion in profit. That list includes Palantir and Tesla, but not your Big Tech players.
Meta, for example, did have to pay some tax last year: an effective federal income tax rate of 3.5% on $79 billion of income in the US, the ITEP said. Just to be clear, the corporate tax rate is 21% in the US. That was achieved, the researchers said, among other ways in part thanks to tax breaks for R&D of $3.9 billion – taxpayers are footing the bill for these companies R&D that will eat their jobs. Talk about a vicious circle. And ITEP noted that Amazon, Alphabet, Meta and Tesla benefited to the tune of $51 billion in federal tax breaks last year alone, after posting $315 billion in US profits. Clearly it pays to attend inaugurations.
Another group, Fair Tax Mark, analysed the tax payments of the so-called Silicon Six – Alphabet/Google, Amazon, Apple, Meta/Facebook, Microsoft and Netflix – over the past decade, finding they averaged a full ten points below what other companies tend to pay.
The report notes that between 2015 and 2024, “the gap between the headline rates of tax and the corporate income cash taxes actually paid was $277.8bn”. That buys a lot of AI research and data centres, though it is over a decade.
If we can’t fix this tax avoidance – totally legal, again just to be clear – then it seems unlikely we can tax our way out of this mess. But we can start by taxing the big six fully now, and taking that money to invest in sovereign AI, unemployment insurance and retraining programmes.
Indeed, there’s another idea floating around, and it’s an old one: redistribution of wealth.
OpenAI CEO Sam Altman – and apparent techno-socialist – wants to give us all AI compute, with the idea we could use or sell it. As yet, my mortgage provider doesn’t take tokens, so I guess I’m taking the cash instead.
OpenAI actually did suggest this in a paper called Industrial Policy for the Intelligence Age, saying AI companies should create a Public Wealth Fund that gives every citizen “a stake in AI-driven economic growth”.
The fund would be used to invest in more AI, but the returns would be “distributed directly to citizens”. Sort of like taxes, but in this case AI companies get to decide with the government how to spend the money on their own industry.
What should we do? All of the above, and then some. Even if AI doesn’t play out quite as disruptively to the labour market as promised, it’s likely going to hit lower-paid people and women the hardest first. So let’s not wait: after all, even if such measures don’t prove necessary in the long run, taking on some or all of these proposals might help make up for the missing billions already owed to society, so it’s a win-win.
