AI might make our work sound smarter, but what happens when we are actually asked to explain it?<\/p>\n
Writing on X on Wednesday, economist John A. List said that while he was initially worried that AI might make critical thinkers \u2014 and therefore economists \u2014 less valuable, his “fears are now assuaged.”<\/p>\n
Having spent the past six months working with nonprofits, corporations, and government agencies, he has watched professionals present AI-generated material<\/a> that sounded polished and persuasive.<\/p>\n List, a Kenneth C. Griffin Distinguished Service Professor of Economics at the University of Chicago and chief economist at Walmart, said the key problem is that they didn’t fully understand what they were presenting.<\/p>\n “The words sound right,” he wrote. “But when someone pushes back just a little bit, the sand castle crumbles.”<\/p>\n Human expertise matters more than ever<\/p>\n List said he observed that AI often produces outputs that are either “very wrong” or “nearly right.”<\/p>\n Spotting the difference requires the same critical thinking skills<\/a> that created the underlying knowledge in the first place, he said.<\/p>\n Being able to defend a conclusion under scrutiny also requires a deep understanding. “It is quite difficult to defend what you didn’t build,” he wrote.<\/p>\n That dynamic, he now believes, strengthens the case for human expertise. The people who can distinguish “nearly right” from “right” are more valuable than ever, he added.<\/p>\n “Creating knowledge still matters,” he said. “Maybe now more than ever.”<\/p>\n Overreliance on AI<\/p>\n AI researchers and some tech CEOs have warned about the risks of overreliance on AI<\/a>, citing, among other things, co-dependency with large language models and a gradual loss of work skills and critical thinking<\/a> over time.<\/p>\n Software developers have had a taste of it when Anthropic’s Claude AI models suffered several outages this week.<\/p>\n