Artificial intelligence is often presented as a growth driver. I argue the opposite risk is emerging.

AI-driven cost-cutting threatens jobs and demand, with the risk of a GDP decline, while shortages of chips, energy, water, and grid capacity threaten higher prices and inflation across the whole economy. Exciting economic policy tools can't resolve either problem.

This video shows why relying on central banks to solve our crises won't work this time, while ignoring unemployment, inequality, and infrastructure will deepen recessionary pressures rather than prevent them.

This is the audio version:

This is the transcript:

I've talked about artificial intelligence many times on this channel.

I've talked about the fact that I think it is going to   cause an economic crash several times.

And every time I look at this issue, I come to the same conclusion, which is that things are worse than I previously thought. And that's not a mistake; that's the result of rational  analysis.

Most commentary that I read on AI is technologically enthusiastic, and maybe that is appropriate.   I'm not arguing that there are going to be no uses for AI; indeed, I've found some. My point is that the analysis of AI in economic terms remains naive. The scale of the risks now accumulating is so significant that I have to make yet another video on this subject, and I genuinely think it's worth your while to hear this one through, because for the first time, I'm going to pull four strands of thought together in a way that suggests AI is really dangerous to our economic well-being right now. All these themes interact. Together, they point to something serious. AI is becoming a source of serious systemic economic risk, is my point.

One of those themes is that financial instability is going to arise from the asset bubble that we know is going on right now, and this is something I've discussed before. There is nothing new in that, but I think that the risk is getting bigger.

The second risk I want to talk about is unemployment risk, and again, this is something I've mentioned before, but it is getting more serious.

The third is a risk to inflation. And I don't think I have discussed this as part of these linked themes before, and so I want to add that into the narrative.

And of course, the fourth is the infrastructure constraints, which I discussed in a video recently, but which now need to be wedded into the common themes that I'm developing here.

So let's look at those four claims and recognise that they do not exist in isolation. They all exist within the political economy of technological change, which, when combined mean that they reinforce each other. That's why we're ending up with systemic risk. They're not isolated. They are literally threatening to drag down the whole of our economy.

So, strand one of my concerns is about asset bubbles.  The global equity markets remain absurdly high. They are at record levels. We've already seen them go beyond the peaks at the end of 2025, in the first week or so of 2026,   and so far, therefore, there has been no global reaction to the overinflated value of AI company shares. But, it remains the case that these valuations assume perfect outcomes for all the investment that is being made in AI, and there is simply very little evidence to suggest that this is going to happen.

The investment that is pouring into  data centres, into chips, and into energy-intensive infrastructure is not sustainable.   I'm going to be exploring that a little later in this video, but the point is, at some point, this is going to be realised, and when it is, it is going to become apparent that much of the current spending on these issues is likely to be wasted. And when we're talking about waste, we're not talking about the odd dollar or pound, here or there. We are talking potentially about  trillions of pounds being wasted, sums so vast that we are discussing the entire income of whole countries   going to waste as a consequence of what is happening. This is why I believe the share crash will happen.

But worse than that, what we now know, and it's becoming increasingly apparent, is that there is a very high degree of risk of contagion when that share crash happens. People are borrowing to buy the shares in AI companies. If the value of those shares falls, the collateral that has been provided for the loans that have been made available to the people who are buying the shares in question will collapse,  and that means that banks will not be able to recover their money, and there will be a market crash and not just an AI share price crash.   And this is just now a question of when and not if.

What is more, it is a matter of when the shadow banking system will also collapse because the shadow banking system has provided half of the money that is likely to have been invested in these shares. And shadow banking is the parallel banking system that exists outside regulation, and it is private equity funds, hedge funds, and so on, who are lenders, but without being subject to banking regulation. They, too, are at enormous risk, and if anything, at greater risk than the banking system itself, because the margins for error there are simply higher because of the lack of regulation. That means that there is massive recessionary pressure likely to arise because of this crash.

But, even without a crash, my point is that this recessionary pressure is going to exist. So far, all commentators agree that  AI is creating increased profit potential for large companies, but not by creating new products. The profit potential comes from cutting costs, and   cutting costs means, primarily, reducing the number of staff who are employed or the pay rates that are paid to them, because the staff in question will be of lower grade than those who will be sacked.

The point is that key people in our society will be made redundant, and that is going to have massive knock-on effects.

It's going to have a knock-on effect with regard to  confidence, and that means that people will save rather than spend.

It's   going to have a knock-on effect because people quite literally will not be in employment, and rising  unemployment is a trend we are already seeing in the UK and maybe in the USA, although there, the data is of lower quality right now because of the government shutdown in late 2025.

What we do know is that overall household demand is falling, and the level of credit to support that demand is rising because incomes are already under threat.

As a consequence, the demand on government spending is going to rise because there will be unemployment, and the multiplier effects within the economy are going to weaken. In other words, there will be less money coming into the economy from people in employment, and therefore there will be less for businesses to receive by way of income and to therefore spend onwards to their own employees and to invest. All of this creates a downward spiral of decline within the economy. And  cost-cutting always results in recessionary pressure, whether it's from AI or anything else that is the outcome, and that is going to give rise to forecasts of stagnation.

We're already hearing of these. People are talking about it. Job losses are expected in the UK now. I've talked about this already this year, and there are serious merchant banks who are claiming that this is the inevitable outcome for our economy in 2026. That AI is risking, deepening recession is really something that we now have to accept as inevitable.

But this has another knock-on effect, and that is that if we do get a recession, government policy should react. The trouble is that government policy on these issues has been outsourced to the Bank of England, and they now have a terrible track record in dealing with these crises.

They missed 2008.

They've been lousy since 2021, and since interest rates rose at that time to supposedly counter the threat of inflation, which had nothing to do with interest rates at all.

And this time I'm expecting them to miss the signs that there will be an increase in pressure on employment, and they will instead look at the pressure of increased borrowing and say that it is exuberance in the market that they're worried about with regard to AI share prices rising rather than the threat to the real economy, which is reflected in increasing unemployment. As a consequence, they're not going to cut interest rates fast enough.

This is where a real policy failure looms. History suggests the Bank of England always reacts too slowly, and I believe they will maintain interest rates far too high for too long, which will deepen any downturn we're going to get.

This is monetary policy failure in action in other words.  Monetary policy is unable to deal with this sort of crisis. It is up to the government to react to a threat to employment, not the Bank of England,   but at present, because we're outsourcing that policy to the Bank of England, we are going to make things worse rather than better as a result of deliberate policy choice.

That, therefore, leads me to my fourth concern, which is the other reason why I think the Bank of England will get this wrong, which is that what we know is that AI is going to stoke inflationary pressure in 2026. This will be seen by the Bank of England in direct contrast to the threats arising from lower employment rates, and they will prefer to tackle inflation even though there is nothing they can do with interest rates to deal with the fact that  a shortage of AI chips around the world is expected to push up the cost of IT this year by up to 20%. This   is a new inflationary pressure imported into the UK,  because we do not make many of these chips, most of them are imported,  and the result will be that prices will rise for consumers in a way that the Bank of England will react to, but which will be absolutely adverse for the economy itself, by keeping interest rates too high.

Couple to that, there will of course be the environmental pressure on  both energy and water, and both of those are going to keep prices high as well, wholly inappropriately,   because somebody's got to pay for the new grid that is going to be required to pay for the increase in electricity demand, which may well be by more  than 10% per annum in the UK, but which will fall on consumers and not the AI companies.

Water usage will also threaten increased water prices, and we're already suffering them.

The result is that we will see inflationary pressure from these sources as well, and the consequence will be twofold. First of all, we will see  increased income inequality in the country as more and more people are forced to spend more and more of their disposable income on essential items   and have less and less to spend on anything else, resulting in that effective reduction in the multiplier effect within the economy as discretionary expenditure falls, therefore leading to recession. Whilst at the same time, the increase in the price of technology will  increase the digital divide in our society because lower-income households will be excluded from access to markets and to government services as a result,   and there is no sign that the government is planning any form of response to this increase in inequality of a practical nature, which will arise as a consequence.

We are therefore having to face a really important question. Why is AI important enough for us to have to go through a recession, an economic crash, an increase in inequality, and the abuse of our planet? What is the justification for that to happen? And I genuinely don't know the answer to that question.

As I've said, I use AI; it's useful, but is it that useful? Is it worth spending this much money on something that we do not even know what it will supply as yet?

Shouldn't we be going slower?

Shouldn't we be working out why we need this thing before we actually throw such money at it?

Should the subscription models that are being used for AI be thoroughly tested, because they currently look fragile, before we hang our hat on them and presume the whole economy can be run on the basis of what they are offering?

The fact is, we don't know the answer to any of those questions.

Like all technologies, AI will reshape part of the economy without doubt, but almost certainly far more slowly than the promoters claim.

My suggestion is simple:  let's go slowly then. Let's walk quietly in the direction   of AI, but let's not crash the economy on the way to doing that because the costs are arriving now, but the benefits aren't, and that's critical. The costs will fall unevenly, and they will fall hardest on those least able to bear them, and that is not accidental. It does, of course, reflect power and its distribution within our society, but this means that we are facing a crisis that is being created because of a failure to sequence events inside the political economy. , Technology is arising before the benefits. The costs are arising before the income. The crisis will happen before we see any upside.

This is a standard story of an economic crisis. It's a story of economic risk gone out of control. It's systemic, and it demands urgent attention because if we don't give it that urgent attention, this coincidence of these four strands of failure, and economic crisis as a result of a stock market bubble, a resulting banking crash, an unemployment crisis, and an affordability crisis, as well as on top of that, an environmental crisis: all of those things will literally lead us to some form of major meltdown. This isn't an incidental; this is an absolute certainty, and it's going to happen, and we need to take action. But where is the talk of this in government planning? Nowhere, and it's that which worries me.

Do our politicians care is my question. Do they care more about AI and the promise of growth than they do about us and the risk to our well-being? It looks as though they do, and that's another reason why our current economic system has to be radically transformed. Our priorities are all wrong. We have to put people first.

What do you think? There's a poll down below.

Poll

 Loading ...

Taking further action

If you want to write a letter to your MP on the issues raised in this blog post, there is a ChatGPT prompt to assist you in doing so, with full instructions, here.

One word of warning, though: please ensure you have the correct MP. ChatGPT can get it wrong.

Comments 

When commenting, please take note of this blog's comment policy, which is available here. Contravening this policy will result in comments being deleted before or after initial publication at the editor's sole discretion and without explanation being required or offered.

PDF of article