2026-06-18T16:00:00+10:00
Scientia Professor Toby Walsh: the pain might be significant but it probably won't last.
Photo: UNSW Centre for Ideas
As AI celebrates its 70th birthday, Scientia Professor Toby Walsh surveys its longer-than-expected history and a few key points about its future.
Artificial intelligence began on the 18th of June 1956. That makes AI 70 years young.
You might be surprised that AI has a particular day on which it starts. Most scientific disciplines don’t start on a specific date. Artificial intelligence is different. Monday, 18 June 1956 was the first day of an eight-week-long workshop whose goal was to kick-start the science of artificial intelligence.
The workshop was held on the leafy campus of Dartmouth College, an Ivy League school in the pretty town of Hanover, New Hampshire. The workshop was organised by John McCarthy, a young assistant professor who I got to know several decades later who had an ambitious dream. His dream was ancient – to build a machine that could think.
McCarthy therefore invited a group of like-minded colleagues from the United States, Canada and the United Kingdom to New Hampshire to join him in building an artificially intelligent future. And he invented the phrase “artificial intelligence” to name what he wanted the group to build.
Funnily enough, 18 June is International Panic Day, which seems appropriate for the day humanity started working on building its successor. Fortunately for humanity, McCarthy’s dream proved harder to turn into reality than he imagined. He had promised the scientific agency funding the 1956 workshop that the group would make significant progress towards building artificial intelligence over the course of that summer. They failed.
So, 70 years later, where are we?
AI models like ChatGPT, Claude and Gemini are definitely very impressive in what they can do. Dario Amodei, CEO and co-founder of Anthropic has talked about having a “country of geniuses in a data centre”. And fields like mathematics and programming are starting to be transformed by these marvellous AI tools.
On the other hand, AI models are also impressive in some of the stupid mistakes they make. Let me give one example with this prompt:
What do you do if your fiancé(e) shows up at your wedding?
Google’s AI Overview offers some confused advice:
If your fiancé(e) shows up at your wedding, it's a situation that calls for immediate and calm communication. The key is to understand why they are there and address any underlying issues. It could be a misunderstanding, a change of heart, or a deeper problem that needs to be resolved…
Artificial intelligence still, then, has some way to go to realise McCarthy’s dream of matching human intelligence. And it’s very hard to say whether that will take a few years. Or a few decades. But I doubt any of my scientific colleagues think it will take a few centuries.
Nevertheless, I’m starting to panic. And not just because it is International Panic Day. What is the impact that AI is going to have on jobs? On our economy? Even on our society?
One common feature of past general-purpose technologies is boom and bust. Railways. Electricity. Internet. All resulted in an investment boom that turned into a bust. And AI is also a GPT, another general purpose technology that appears to be repeating that cycle.
Why is it that we can never learn from history? Billions of dollars are being invested into AI every day. And many people are starting to wonder when or if we will see adequate financial returns for this unprecedented scale of investment.
I fear that the current AI boom will end in an AI bust. This will, at best, transfer vast amounts of wealth from the investors into artificial intelligence to the small number of AI companies that survive the bust.
The good news is that, when that bubble bursts, history suggests the pain won’t last. We will get to an AI future with the gains being promised today. Just as we eventually got to the internet economy promised before the dot com bust.
Let me make one other prediction. Sometime in the future, after the AI bubble bursts, and after artificial intelligence emerges that matches, perhaps even exceeds, human intelligence, there won’t be just one winner.
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 the artificial intelligence market won’t, either, be won by a single player.
Intelligence wasn’t a winner-take-all market. There are 8 billion intelligences on the planet today. And there will be eight billion artificial intelligences in the not-too-distant future. You’ll have one that runs on your smartphone – having it run on your own device will give you privacy and many other advantages such as cost and latency.
In the long term, the AI future, then, isn’t the one that Silicon Valley is currently trying to sell you in the short term. It’s a more democratic and affordable one. But someone is going to have to pay the price for all those data centres.
Toby Walsh is Professor of AI and Chief Scientist of UNSW.ai, the AI Institute at UNSW Sydney. He is author of several books on AI for the general audience, including “The Shortest History of AI”.
For enquiries about this article or to arrange an interview, please contact UNSW Media.
Tel: +61 2 9385 2864
Email: media@unsw.edu.au
UNSW
Sydney NSW 2052 Australia
Telephone: +61 2 93851000
UNSW CRICOS Provider Code: 00098G
TEQSA Provider ID: PRV12055
ABN: 57 195 873 179
UNSW respectfully acknowledges the Bidjigal, Biripai, Dharug, Gadigal, Gumbaynggirr, Ngunnawal and Wiradjuri peoples, on whose unceded lands we are privileged to learn, teach and work. We honour the Elders of these Nations, past and present, and recognise the broader Nations with whom we walk together. UNSW acknowledges the enduring connection of Aboriginal and Torres Strait Islander peoples to culture, community and Country.
The Uluru Statement
Leave a Reply