Better stop pretending that job losses from AI are no big deal

Image credit: Zenzen /

One of the many great debates over artificial intelligence is the impact on jobs. It’s already well established that AI can enable sufficient automation to take over certain types of low and mid-level jobs. But that doesn’t necessarily mean mass unemployment – some analysts say AI will change the nature of the job rather than replace the job itself, while others say AI will create more jobs than it kills. So, no need to worry, right?

Well, according to Kai-Fu Lee – founder and CEO of Sinovation Ventures and president of the Sinovation Ventures Artificial Intelligence Institute – you probably should worry, yes. Artificial intelligence constitutes a major transition in the workforce, and we’re nowhere close to ready for it.

In a recent column for MIT Technology Review, Lee argues that tech companies are kidding themselves if they think that AI won’t destroy jobs permanently. Now that AI is essentially an arms race between China and the US (and one that China has a 2:1 chance of winning at this stage), AI development is happening at such a furious pace that the impact is going to catch us off-guard:

The rise of China as an AI superpower isn’t a big deal just for China. The competition between the US and China has sparked intense advances in AI that will be impossible to stop anywhere. The change will be massive, and not all of it good.

The problem is that artificial intelligence has the potential to displace tens of millions of blue-collar and white-collar workers, which could result in serious social, political and economic consequences. We already live in uncertain times as various traditional paradigms and power structures are being decentralized and disrupted on various fronts. Putting millions of people out of work won’t help stabilize things. A recent poll from Northeastern University in the US indicates many workers already believe that AI will take over lots of jobs – and they’re understandably worried about that.

Lee isn’t impressed with the optimists – it’s wrong to assume those new jobs will materialize, or that displaced workers can be easily trained to take up those jobs. As for AI-assisted jobs, those will be vastly outnumbered by single-task jobs that AI can take over permanently.

It’s also disingenuous to assume that businesses will do everything in their power to make sure workers aren’t displaced. Maybe some will. But businesses already lay off workers by the tens of thousands in the name of efficiency and cost-cutting – it’s a standard business practice. If they’re presented with a technology that enables them to save millions in operating costs – and the cost of implementation isn’t too steep – odds are they’re going to take it.

Lee is especially dismissive of the argument that the impact of artificial intelligence on jobs could be mitigated by a universal basic income funded by the extra money made by AI. Ha ha. No:

… UBI doesn’t address people’s loss of dignity or meet their need to feel useful. It’s just a convenient way for a beneficiary of the AI revolution to sit back and do nothing.

Aside from that, good luck convincing businesses to give up their rising profits to fund a UBI. And good luck getting a UBI approved in certain countries where certain political parties are practically allergic to anything that looks like it might be something that vaguely resembles that Horrible Evil Thing That Shall Not Be Named But Starts With ‘S’ and Ends With ‘Ism’.

None of this is an argument against AI, of course. Aside from the fact that AI development is a fait accompli, the potential benefits are substantial enough to justify pursuing. And it’s always possible that the impact of AI on jobs is exaggerated – after all, we’ve heard doomsday predictions about the impact of automation on the workforce many times before.

On the other hand, the difference this time is the speedier rate of change and the broader scope of jobs being affected across multiple industries. Thus, Lee argues, it’s dangerous to assume AI will play out the same way.

Tech companies – and all businesses, really – need to take this far more seriously, and they need to take action now to minimize the disruptive impact, he says:

These changes are coming, and we need to tell the truth and the whole truth. We need to find the jobs that AI can’t do and train people to do them. We need to reinvent education.

Easier said than done, obviously. But the place to start is to stop pretending job losses resulting from AI won’t be a problem. We’re better off assuming it will be.


  1. Of course if socialist governments want to be creative, they could tax the work that Robots do as they would any particular individual, or charge companies an AI tax for any task that AI innovation does. And at the end of the day, what good is AI if people do not have jobs or income to buy the products or services that AI create?

  2. Not clear why ‘Tech companies – and all businesses, really – need to take this far more seriously’. As you say elsewhere, they’ll continue to do their thing – making profits wherever and however they can. It’s governments that need to worry about the effect on the populace and the broader economy…
    This isn’t really to do with AI, is it, or only peripherally? It’s more to do with the onward march of automation. Where workers can be replaced with machines or software, they will be, and the recent rise of the gig economy will make many all too easy to replace. As to how many will benefit, I’ve been hearing all my career that IT (another word for automation) will ‘free people up to do more interesting and fulfilling jobs’ and rarely seeing that come to pass (unless it means working as a prototype robot at your local Amazon ‘fulfillment centre’). It’s possible that when it all shakes out all will be well (and I increasingly think some form of UBI will be part of that) – but history suggests it will take two or three difficult generations to happen.

  3. Sorry, but this article does not make convincing reading, despite having read through it carefully two times. The core of the argument – that AI will replace “millions of jobs” needs to be more carefully demonstrated through facts, some logic, an application of how people think in real life and research, notwithstanding polls and research from AI experts. As an example from a website that I have no connection or involvement with, it may be worth reading up on adversarial research in AI:

    The article even goes on to say “And it’s always possible that the impact of AI on jobs is exaggerated”, which means that the article is backing away from its title. It would be better if there were cogent arguments that are plausible in the real world rather than saying either outcome (or a mixture) is possible. How does that shine light on the topic instead of spreading heat?

    The passing comment about socialism and a UBI does not help bring healthy debate to the AI topic and starts to step into political debate. In this of course people are free to express their opinions (and not be silenced) but on the topic of politics the opinion expressed just becomes another person’s opinion amongst the millions available today and detracts from the expertise and industry knowledge that the writer does have.

What do you think?

This site uses Akismet to reduce spam. Learn how your comment data is processed.