Technologists used to be cheery types — often boosterish to the point of annoying. So it has been a little disorientating lately to watch a parade of tech leaders issue public warnings about the dire potential consequences of their own inventions.
Mustafa Suleyman, one of the co-founders of AI lab DeepMind, told an event in San Francisco this month that AI would threaten white-collar workers and create “a serious number of losers” who would be “very unhappy, very agitated.” Governments, he said, would have to think about how to compensate them.
Sam Altman, chief executive of OpenAI, was encouraging last week in testimony to Congress about the potential for new work to be created, but nonetheless said there would be some “impact on jobs” which would “require partnership between the industry and government, but mostly action by government, to figure out how we want to mitigate that.”
Good for them, you might say. Better to be honest about the potential risks automation poses to some workers than to be blindly optimistic. After all, everyone remembers learning about the Luddites in early 19th century England, who were so angry they resorted to smashing the weaving machines that had displaced them. I would agree with that. But I also think there is something dangerous about the narrative that AI will create winners and losers, and that the losers must be compensated.
It gives the impression that the outcome of technological change is inevitable and all we can do is get governments to mop up the aftermath. But how AI changes the world of work is far from pre-determined: it will depend on the balance of power in millions of different workplaces; on legislation and regulatory enforcement; on the outcome of fights over ideas, over laws, over working conditions and over the distribution of productivity gains.
As economists Daron Acemoglu and Simon Johnson argue in their new book Power and Progress, this has been the case throughout history, from agriculture in the middle ages to the factories of the industrial revolution.
The Luddites are actually a good example. They weren’t against technology itself so much as the way it was being implemented to undercut their skills and replace them with shoddy products made by unskilled workers or children. They turned to machine-breaking in desperation, but their demands weren’t unreasonable to 21st century ears: their ideas included minimum wages, minimum labour standards and a tax on machine owners to help support unemployed workers. They just didn’t have the power to get anywhere.
There are echoes of those fears about undercutting and deskilling in the fight under way today in Hollywood, between writers and studios over how AI could be used in the scriptwriting process. Backed by strong unions, the writers might be able to negotiate a fairer way forward.
The other problem with the “compensate the losers” narrative is that it makes it sound as if government is the only solution. This lets companies off the hook, and it also implies individuals have no agency.
A new policy in Sweden is an interesting example of an alternative approach: the country has created what is effectively a furlough-type scheme for life-long learning. Under an agreement between employers, unions and government, workers can take time off to train in something new, while being paid 80 per cent of their salary (up to a limit).
This isn’t waiting until people are made redundant then offering them a cheap course so you can tick the “retraining” box in the “compensating losers” handbook (which didn’t go so well in the deindustrialised areas of many developed countries in recent decades).
Instead, the idea is to be proactive — to help workers stay ahead of changes in the world of work, and to help employers by raising the skills base of the workforce. It is still early days and there are some dangers, for example that it displaces employer-provided training.
But Swedish unions hope it will “make our members safer in the labour market and more resistant to destructive forces that are always at work in a small open economy,” says Fredrik Söderqvist, an economist for the LO — the Swedish trade union confederation. “This sort of highlights a basic tenet of the Swedish model — security in the labour market is supposed to bring security in the individual job — not the other way around.”
It’s time to stop saying AI will produce winners and losers, as if the whole thing is out of our hands. It creates opportunities and dangers. How they play out is up to us.