A.I. will reduce inequality by leveling the tech playing field, studies suggest
Good morning.
Much of the commentary around generative A.I.—including my own—has focused on negative implications for society. I’ve been particularly disturbed by the tendency of large language models to just make stuff up—or “hallucinate”—occurrences that technology experts tell me are more feature than bug.
This morning, some good news. Economists generally agree that technology has fed inequality in advanced societies over recent decades, putting outsized power and money in the hands of math nerds, computer geeks, and other over-educated experts. But there’s emerging evidence that generative A.I. may be the first computer technology wave that has the opposite effects.
A spate of recent studies, summarized last week by the perceptive blogger Noah Smith, suggests generative A.I. gives its biggest boost to the lowest skilled. A study by Erik Brynjolfsson and two of his colleagues gave generative A.I. tools to customer support staff and found it “disproportionately increases the performance of less skilled and less experienced workers.” Another study, which gave the tools to professionals to help with writing tasks, found that productivity inequalities among the group “were half erased” by the technology. Yet another looked at the effect of GitHub’s “Copilot” on programmers and concluded “less experienced programmers benefit more.” Still another, looking at how generative A.I. could help students on law exams, found “students at the bottom of the class saw huge performance gains with A.I. assistance, while students at the top of the class saw performance declines.”
This is a big deal. It suggests that while earlier waves of information technology complemented human skills, generative A.I. tends to substitute for such skills. Smith uses the analogy of a shovel—which increased the power of able-bodied workers—versus a steam shovel—which did the work for them. It’s still early days, of course, but there’s at least some early evidence, as Smith puts it, that generative A.I. “could level the human-capital playing field.”
That’s one of the topics we’ll be discussing Oct. 3, when the Fortune CEO Initiative meets in Washington. Brynjolfsson will be joining us, along with tech savvy CEOs like IBM’s Arvind Krishna and Accenture’s Julie Sweet. You can read Noah Smith’s report here and get more information on the CEO Initiative here.
Here’s another topic we’ll be discussing at the CEO Initiative: How the Supreme Court’s decision on affirmative action is affecting corporate diversity efforts. A study by Bloomberg found that mentions of diversity, equity and inclusion on earnings calls and at conferences fell 54% in the third quarter from a year earlier—before the court’s decision. Still to be determined: Whether companies are just talking less or actually doing less. Former Merck CEO Ken Frazier, who has been a relentless champion of corporate diversity efforts, will be with us in Washington. I also plan to raise the subject on Wednesday this week when I interview politician and activist Stacey Abrams at the Fortune Impact Initiative in Atlanta.