Opinion: Artificial intelligence may transform the economy over the long term, but investors betting on today’s AI boom should remember the lessons of railways, dotcoms and every great technological mania before them, writes Toby Walsh.
History, it is said, tends to repeat itself, especially for those who fail to learn from the lessons of the past. And when I look at the AI race today, I fear that we have been here before and are failing to do exactly that: learn the lessons of the past.
At the start of the industrial revolution spending on railroads eventually peaked at around 20 per cent of US GDP. Spending on artificial intelligence is already a few percent of US GDP and is set to rise even further. Indeed, spending on AI is one of the few things preventing Trump’s tariffs and a misguided war in Iran driving the United States into a recession.
But don’t forget how all that spending on railways ended. It made a few robber barons like Cornelius Vanderbiltand Leland Stanford. But most investors lost everything.
The severe international financial crisis that followed, known as the Panic of 1873, bankrupted countless small-time investors who had put their hard-earned money into railway bonds. History was repeated in the Panic of 1893. This produced the most serious economic depression in history until the Great Depression of the 1930s.
The great wealth transfer
But perhaps those two panics are too far back in time to seem relevant today. So, what about the dotcom bubble?
From 1995 to March 2000, the value of technology stocks rose six-fold. When that technology-driven bubble burst, the Nasdaq gave up all its gains. The market lost over US$5 trillion in value, with numerous companies facing bankruptcy and contributing to a recession.
Of course, the modern internet economy emerged from the ashes of this bubble bursting. But the internet bubble transferred vast amounts of wealth from speculators and early-stage investors to the companies and entrepreneurs who survived and could scale during the recovery,
The current AI boom will likely do the same. It will transfer vast amounts of wealth from those investors in artificial intelligence to the small number of companies that win the AI race.
And with many of the AI companies going public and joining the indices that superannuation funds track, we, the public, are now those exact same investors who might stand to lose.
Will AI Deliver Returns?
There’s a saying attributed to Roy Amara, long-time head of the Institute for the Future that we tend to over-estimate the impact of a new technology in the short term, but under-estimate its impact in the long term. This seems painfully true of artificial intelligence.
I suspect the AI companies themselves are making exactly this mistake. They are spending vast sums building data centres, training models and acquiring talent but doing it much too rapidly, as they over-estimate how quickly AI is going to give economic return. Artificial intelligence will likely deliver trillions of dollars of value over the coming decades. The mistake is assuming those returns will arrive in a year or two rather than a decade or two.
The expectations of the impact of the technology are great. Dario Amodei, the CEO of Anthropic has, for example, predicted half of all white-collar jobs might go away. Demis Hassabis, the CEO of Google DeepMind, expects it to supercharge science.
But we’ll likely only see such gigantic impacts in the long term. And we’re burning cash at a rate today that requires those long-term impacts to arrive in the short term.
There’s only one way this ends. With the AI bubble bursting. But again, we should learn from history that the pain won’t last. We will, I’m sure, eventually get to an AI future justifying the scale of investment being made today.
There will be more than one AI winner
Let me make one other prediction. Sometime in the future, after the AI bubble bursts, and when artificial intelligence emerges that matches, perhaps even exceeds, human intelligence, there won’t be just one winner.
Now, there are many winner-take-all races. Search. Social media. Ride-sharing. But no general purpose technology so far has proved to be a winner-take-all market. And I’m confident that artificial intelligence won’t be one either.
Intelligence itself was never a winner-take-all market. There are eight billion intelligences on the planet today. In time, there may be billions of artificial intelligences too, embedded in our businesses, homes, devices and daily lives.
Historians will, I suspect, look back at the AI bubble bursting as the catalyst for that artificially intelligent future. Just don’t bet your pension on it in the meantime.
Toby Walsh is Professor of AI at UNSW Sydney and Chief Scientist of their AI Institute. He is author of several books on AI including most recently, “The Shortest History of AI”.
Want to see more Forbes articles on your feed? Tap here to make Forbes Australia a preferred source on Google.
Look back on the week that was with hand-picked articles from Australia and around the world. Sign up to the Forbes Australia newsletter here or become a member here.
