Sociologist Zeynep Tufekci argues in this New York Times column that fears of mass AI-driven job loss are overstated. She describes large language models as “plausibility engines” rather than reliable reasoning systems, citing failures such as Air Canada’s chatbot wrongly promising a refund and McDonald’s abandoning a drive-through AI bot after it added large erroneous orders. She concludes that only formal, easily verifiable work such as coding is likely to be readily automated, while most jobs requiring judgment will remain dependent on humans.
