Studies Suggest That Rather Than Killing Jobs, AI Could Revive The Middle Class

We've certainly been talking a lot about the AI Doomers" who insist that AI is all too likely to destroy humanity. However, even people who aren't fully on board with the existential threat of AI do often say that, at the very least, it's going to destroy jobs for most people, potentially creating huge problems. For years now, people have been arguing for universal basic income, in large part, because they think that automation and AI will take away everyone's jobs. I mean, it was a core plank of Andrew Yang's silly run for President.

And with all these new AI tools coming on the market, there are discussions about it again. In recent weeks Annie Lowery in the Atlantic and Andrew Kennedy in Newsweek, both argued that AI was going to wipe out millions of jobs and we needed UBI to protect against that future. Both articles were responding to a Goldman Sachs report predicting massive growth in GDP from AI, but also that 300 million jobs might be automated away.

Of course, both of those articles totally ignore that the very same Goldman Sachs report predicts that the automation of those 300 millions jobs might not lead to mass layoffs, but rather just a changing of how those jobs are done, saying that in many cases, it will complement jobs, rather than substitute for it.

And, a few early studies seem to support that possibility. Planet Money recently had an episode looking at these new studies that even suggest that these new AI tools could help revive the middle class.

But there is a glimmer of hope - in the form of an economic study. The study looked at the customer service department of a big software company, and it found that ChatGPT made workers much more productive. More interesting, most of those gains came from less skilled workers, while the more skilled workers showed only marginal improvement. Put in other words, AI narrowed the productivity gap between lower skilled workers and workers with more skills. This finding is very different from previous findings about the effect of technology on workers over the last four decades. A whole generation of economic research shows that computers have been a major force for increasing inequality. A force for a shrinking middle class.

David Autor is a professor at MIT, and he is widely regarded as one of the greatest labor economists in the world. He led a lot of that initial research about the computer era and the labor market. And he thinks this study, and another one like it, suggest that we could use AI to expand job opportunities, lower barriers to entry to a whole range of occupations, and reduce inequality.

It's notable that Autor is saying this as well, since (as the article snippet notes) his earlier research has shown how earlier tech innovations actually increased inequality. His biggest papers are on how much tech has harmed inequality in the job market. But in the podcast, he notes that his own research suggests that AI may also be decreasing that inequality.

The key, again, is recognizing how AI works best as a tool to assist workers, rather than as a tool to replace them. The first study mentioned above is really interesting to read. Basically, they looked at how customer support agents were using AI, and specifically reviewing productivity and performance and how it changed via the staggered introduction of the tools to different support agents.

The report found productivity went up across the board - and also that both employees and customers were much happier, which seems like a good thing. But, the impact on lower skilled workers was much bigger, basically leveling them up to work nearly as well as higher skilled workers.

Instead of experienced and skilled workers benefiting mostly from AI technology, it's the opposite. It's the less experienced and less skilled workers who benefit the most. In this customer support center, AI improved the know-how and intelligence of those who were new at the job and those who were lower performers. It suggests that AI could benefit those who were left behind in the previous technological era.

And that might be helpful in terms of closing some of the inequality that previous technologies actually helped amplify," Brynjolfsson says. So one benefit of intelligence machines is - maybe - they will improve the know-how and smarts of low performers, thereby reducing inequality.

Of course, this is just one study of one company, using the technology as it is today. But lots of things could still change. That study may not prove to be generally applicable. The tech could change. There are lots of other things that could go wrong. And this isn't to suggest that UBI might not still be a useful tool for helping make sure that everyone has a clear foundation on which to feel safe and able to live freely.

But, at the very least, it might call into question the doom and gloom" predictions that this will somehow further hollow out the middle class and massively increase unemployment. The early evidence, at least, seems to suggest it might do the exact opposite.