Will Humans Still Be Needed in the Age of AI?

Summary

The article examines whether artificial intelligence will merely automate tasks or eventually make large groups of people economically and politically unnecessary. It begins with scenarios from Anthropic economists for the US economy in 2030: GDP could rise by roughly one third while employment falls substantially, although the scenarios are explicitly not forecasts or probability estimates. The author argues that predictions about rapid robotization have repeatedly failed, citing autonomous cars, Baxter, and Geoffrey Hinton’s forecast that AI would replace radiologists; in each case, technical limits and expanding demand complicated simple displacement stories. Historical examples such as radiology, paralegal work, bank tellers and programming are used to explain Jevons-like effects, in which lower costs increase total demand and automation changes tasks rather than eliminating professions. The author nevertheless argues that this wave could differ because large language models claim broader intellectual generality, while autonomous driving, humanoid robots and AI-enabled military systems are also advancing. Waymo and Baidu’s Apollo Go are presented as evidence that robotaxis are scaling, even though their services remain geographically limited; the article also notes declining bank-teller numbers after mobile banking and the renewed development of humanoid robots. The proposed test for whether people remain needed has three parts: military usefulness, labor value and humanitarian or social value. Humans still provide control, resourcefulness, judgment, taste and intuition in areas where AI remains uneven, while drones already account for much of the casualty burden in the war in Ukraine and AI systems are increasingly relevant to speed-sensitive operations. The article concludes that the political outcome will depend heavily on who owns the AI and robots, whether human capabilities such as originality and judgment retain economic value, and whether automation produces broad prosperity or concentrated power. A basic income could become feasible if automated systems produce abundant goods, but the author treats that as a political choice rather than an automatic consequence of technological progress.