Last week the NY Times ran an article, by a M.I.T. economist and “tech contrarian” David Autor.  In the post entitled How One Tech Skeptic Decided A.I. Might Benefit the Middle Class,  Autor puts up a case for A.I. not leading to mass joblessness and perhaps even boosting those middle class.  

I don’t even think he believes his own argument. 

One of the reasons he provides is throughout economic history new technologies replace old ones demanding new kinds of jobs and expertise, but what happens when our jobs are replaced by expertise itself?

He contends that AI will be a great partner to workers, giving us the ability to take higher value jobs, improving job quality and pay for what he says are non-college educated workers. That means less jobs will require a degree. How will colleges adapt?  

Autor admits AI will automate some tasks, devalue some human expertise, and it is up to the institutions and companies who own the processes to ensure humans are equally valued.  Sure A.I. does have the potential to improve job quality and compliment our expertise in the field, but he is overly optimistic that institutions will feel the least bit responsible to choose person over productivity. 

If the rate of adoption and competition continues the middle class will shrink and perhaps the only thing that will actually bolster it would  be “universal” income.  The possibility is there that we can be so productive with new tools we’ll need a new way to finance this country, but that’ getting ahead of myself. 

If you accept Autor’s argument that AI will create as many or more jobs than it destroys, there is no guarantee that the new jobs will be good ones that support a stable middle class. Autor admits  one possible scenario is a world in which a small group of “owners of Artificial Intelligence patents” capture most of the wealth while the rest of humanity is left with low-value work. Like we automated manufacturing more that a century ago, with this we are automating all work.  To keep people employed we will need more manufacturing and service jobs. That will mean this country will need to change how it thinks about work.  AI can’t flip burgers but can run a hedge fund. 

There is no slowing down the rate of AI adoption.  The biggest, most powerful companies are operating in an Arms Race to see how fast they can adopt.  Google’s Ad division has laid off hundreds.  Ikea replaced their entire call center with an AI bot.  Salesforce, Duolingo have already laid off hundreds and IBM plans to replace 30% of their back-office roles with AI.  It’s scary how fast everyone was able to release their AI product once OpenAI launched ChatGPT.  Were they all slow in launching their own product because they wanted to get it right and ethical but panicked.  

We’ve seen technologies come and “hollow out” the middle-skill jobs like bank tellers and now AI is supercharging the trend and coming for high-skill professions like coders and accountants. Now theirWe’ve already seen how recent technologies have contributed to a “hollowing out” of middle-skill jobs and a polarization of the labor market into low-wage service jobs on one end and high-skilled professional jobs on the other. AI can only supercharge this trend.

There is no track record that the money behind AI, like tech giants and VC backed startups care about using AI in a way that benefits works.  There is a major power shift happening. We’ve already seen AI systems exhibit racial and gender bias, and be used to enable mass surveillance and manipulation. Human workers are making less decisions in decision-making contexts from hiring to medical diagnosis. The corporations and governments deploying these systems are doing so to cut costs, centralize control and accumulate power.

Autor is right that “AI poses a real risk to labor markets, but not that of a technologically jobless future,” I worry he is simply waving away the jobless future risk too quickly. He largely treats the development of AI as a given, something that will just happen to us one way or another. But in fact, we have agency in shaping what AI technologies are developed, how they are deployed and who benefits. Leaving it up to the “free market” is a choice, not an inevitability.

We desperately need much more proactive government policy, stronger worker and consumer protections.  Also we need reforms to intellectual property law as AI exploits human expertise and creative forces. We need a serious discussion about more transformative solutions if AI does end up displacing labor on a mass scale. Those could include policies like universal basic income and robot taxes. 

Autor’s impulse to sketch an optimistic vision hand waves away the challenges and puts too much faith in the automatic beneficence of market forces. The fact that human expertise has found a way to evolve and stay relevant after past technological revolutions doesn’t guarantee the same will happen with AI. To be clear, I agree with Autor that the path he describes.is absolutely one we should strive for. I’m not convinced it will happen without a concerted effort to steer AI towards more governance. 

As Autor writes, “This alternative path is not an inevitable or intrinsic consequence of AI development. It is, however, technologically plausible, economically coherent and morally compelling.” Faith won’t cut it. We need a say in the development of AI for the common good.