Could artificial intelligence trigger another Great Depression? And could the very success of AI cause the next global financial crash?

AI investment is booming, technology companies have extraordinary valuations, and financial markets are betting on enormous future profits. But what happens if those profits never materialise?

In this video, I explore an argument made by American economist Professor James Galbraith, whose interpretation of the Wall Street Crash of 1929 offers a disturbing warning about today's economy.

Galbraith argues that the discovery of cheap American oil undermined the profitability of industries built around coal, steam and railways. Technological progress offered enormous opportunities, but destroyed the financial expectations underpinning the existing economy. The resulting financial crisis prevented investment in the new infrastructure required.

Could AI create a similar crisis?

American technology companies are investing hundreds of billions in AI. Their valuations depend on enormous future profits. But what if China develops better AI at a fraction of the cost?

American investments could become worthless. Share prices could collapse, pension funds and banks could suffer enormous losses, and financial instability could spread worldwide.

AI could succeed technologically while causing financial disaster.

This exposes the dangers of shareholder value and financial markets that prioritise speculative profits over economic progress.

I explain why Roosevelt's New Deal, which dragged the USA out of the 1929 crash, offers lessons for today. Government investment helped America make the transition to an oil-based economy. Similar public intervention may now be essential to ensure AI benefits everyone.

The internet survived the dot-com crash. AI could survive another financial crisis. But could the global economy?

Galbraith's warning is that governments must act before another crash, not afterwards. The question is, will they?

This is the audio version:

This is the transcript:

Could AI cause another Great Depression? That's the question I need to ask in this video because the possibility is very real. AI investment is booming. Financial markets are expecting enormous profits to arise from AI activity. But what if AI succeeds too well? What if it succeeds so well in China, for example, that we get a financial crash anyway because our confidence in AI has been misplaced as to where it will come from? Could it be that the financial system and AI could be in conflict with each other and we get a crash? History suggests that this is a possibility.

The evidence is that technological success can cause financial failure. New technologies can threaten existing investments, and as a consequence, the expected profits from them can disappear overnight, and that can then cause the financial crash, which spills over into the rest of the economy.

Financial markets might then collapse, and that can happen even when the real economy could actually be improving, but financial markets pull it down. This is the contradiction we must understand. And James Galbraith, a professor of economics in the USA, has offered a surprising explanation for this phenomenon based upon his interpretation of the Great Crash of 1929.

At that time, America discovered enormous quantities of domestic cheap oil that changed its economic prospects. Coal, steam and railways had been the foundation of wealth in the USA to that point. Now, oil faced the prospect of disrupting all of that. All existing investments, particularly those in railways and the type of manufacturing that fed into that system, suddenly looked as though they were irrelevant.

They collapsed. Their future profit expectations disappeared, and the consequence was the crash of 1929. It was not perhaps as irrational as it looked. The discovery of oil shortly before that crash changed expectations of profitability. The reaction in terms of a crash was rational. The economy could still function, but the financial returns that it was expected to generate were now entirely different. And the problem was very clear.

Whilst the old economy could still work, the financial system which it had supported could not anymore because the profit expectations on which it had been built had now disappeared. But the investment to build the new infrastructure that the oil economy required was not available because financial markets were collapsing.

The problem was that the old system was collapsing in terms of future profit expectations and, as a consequence, was disrupting financial markets. And as a result, those financial markets didn't have the finance available to fund the new. The USA was stuck in the middle in 1929. The steam- and coal-based economy was collapsing. The oil economy in the USA was trying to emerge, but the finance was not available to give rise to the new.

That is what James Galbraith has said happened then. It's a new interpretation. I find it quite compelling. I think there's a lot of truth to that. And the great appeal of this explanation is it is the real economy that explains the change.

The financial system collapsed because of a real-world event. It didn't just collapse because of economic madness or overpricing. It collapsed because expectations were destroyed by a real-world change.

Now let's be clear as well: that real-world change created enormous potential inside the USA. The oil-based economy, as we know inside the USA, became the basis for the post-war reconstruction, not only of that country, but many others as well. The potential within this discovery was sufficient to create enormous well-being.

The problem was that existing companies did not have the means to adapt. The banking system, which had lost its money investing in the old system, could not provide the finance that was required. And the transition was therefore one which was in itself unsustainable. This is where the problem arose. The financial system could not manage the transition to the new economy that the USA required.

And AI could repeat this pattern. That is what James Galbraith is warning. American technology companies are investing billions, many, many billions into AI. Their valuation does, of course, depend upon future profits, and those profits are far from certain. The question that James Galbraith asks is: what would happen if China could produce much cheaper, and at least as good, AI as the USA can?

If that happens, American investments could suddenly become profoundly uneconomic. The rate of return on all that money being spent now would basically disappear. China could sweep the board with cheaper and more efficient AI than America can offer. The consequence would be the requirement for another transition to a new economy, an AI-based economy, but all the money that had been invested in its creation in the USA would basically prove to be worthless.

In that case, AI's very success could actually trigger a financial collapse. The success would flow to China, but there would be a benefit to users. They would get access to cheaper, better systems as a consequence, and some people say that is what China is already producing. But American technology and its profits could disappear.

Share prices would fall dramatically. Pension funds and banks would suffer. We would get the crash that I've talked about many times before on this channel. Financial losses could then spread throughout the whole economy. And all of this suggests that there is a real problem with organising the economy around shareholder value.

The investment in AI in the USA embodies the idea of shareholder value. A very few companies, led by a very few individuals who have become rich beyond imagination are typifying the idea of shareholder value and the expected returns that are implicit within it. Everything about shareholder value is about capitalising future returns to create worth at the moment. This is what creates the value of these people. They haven't got the money yet. That has not happened. But the expectation of markets is that these AI companies will create such large profits in the future that they have enormous value now.

But that does not prove that they have yet to create a valuable business model. And if somebody else, the Chinese in particular, could produce technology that undermines this model, the idea around shareholder value that drives the concentrated investment in just a few companies in the USA could be blown apart. Financial interests could then be the obstruction to genuine economic progress in the US economy. And that contradiction now threatens economic stability.

So how would America have to get out of this situation? How would the world have to get out of this situation? The lesson also comes from 1929 and its aftermath. After all, how did America escape the Great Depression?

Markets most certainly did not drag the USA out of that depression in the 1930s. It was Roosevelt and the New Deal, launched in 1933, that did that. Government invested in the essential infrastructure to make the oil age happen.

It built roads. Those were essential. Cars couldn't use this oil unless there were roads to drive on.

It built the facilities to transport oil.

It transformed its railways from a steam-based system to a diesel-based system long before anything like that happened in the UK.

And public investment helped create the new opportunities that the oil age offered, including, of course, the airports that did eventually drive us all into the air.

Markets did not deliver this transition. They were, of course, involved, but it was government investment that crowded in private investment that made the oil age possible. Coal and steam died in 1929. The oil age began with the government investment driven by Roosevelt's New Deal in 1933.

And what does that tell us now? Well, what it does say is we shouldn't leave it too late. If AI creates risks, and we do all know it creates risks, the requirement is that the government must act now. AI cannot be left to corporations driven by shareholder value to develop and direct what it does. That is not an outcome which is likely to produce public value.

Public investment must instead support this process of technological change and help direct it to social purpose. That is what is happening in China. That is not what is happening in the USA. We have to ensure that there is a solid base from which AI serves the public interest. And financial speculation must, in that case, be controlled because it threatens that public interest outcome. Real resources and not money must determine our possibilities and the balance of risks, and at the moment, we're putting money ahead of everything else when it comes to AI. We are therefore creating a boom that could lead to a bust.

We could avoid that depression. That's what James Galbraith is saying, and I agree with him. But that requires AI to be a success for everyone, and everyone must benefit. Now I think there are ways in which that is possible. I do not doubt that there are benefits to be had from AI, just as in 2000, there were benefits for everyone to come from the internet. But at that moment when we had the dot-com bubble burst, it was not clear where those benefits were coming from. It is not yet clear whether benefits from AI are coming from, and that is precisely why we need public involvement in this process.

Financial markets must not, and even cannot, dictate technological progress. Shareholder value cannot be our economic objective. We must be working on something as radical as AI in the public interest. And so government must be involved in managing this process of change for society. That's not just in the USA. That's in the UK as well, in the EU and elsewhere.

But the question is, will governments act before another crisis? That is what James Galbraith is asking, and it's a good question at this moment. The risk is, candidly, that China, with its very different approach to AI, which is on a much lower investment base, but with a goal of achieving a level playing field outcome, might just be ahead of the game in reality. And if they are, we're going to see the most almighty crash as a consequence. I have a feeling that James Galbraith's hypothesis on this might be horribly right.

That's what I think. What do you think? There is, of course, a poll down below. Let us have your comments. Please do like and share this video. Please do subscribe to our channel. And if you like what we do and would like to support our work, there is a link down below to our Ko-fi page where you can make a donation. Thank you for your support.

Poll

PDF of article