Technology Review is reporting that Nobel laureate Daron Acemoglu maintains his measured perspective on artificial intelligence's impact on jobs and productivity, despite growing public concern. The publication's "The Algorithm" newsletter highlighted Acemoglu's 2024 paper, which estimated AI would provide only a small boost to U.S. productivity and would not eliminate the need for human work. It said that while fears of an "AI jobs apocalypse" are widespread, current data continues to show AI is not significantly affecting employment rates or layoffs.
Acemoglu discussed agentic AI, tools capable of operating independently to complete goals. He believes these agents are best viewed as tools to augment specific tasks rather than replacing entire human jobs, citing the complex orchestration of various tasks humans perform naturally. He noted that many jobs would be spared from an AI takeover if agents cannot fluidly switch between diverse responsibilities.
The newsletter also reported on a trend of AI companies, including OpenAI, Anthropic and Google DeepMind, hiring in-house economics teams. Technology Review said Acemoglu has observed colleagues taking these roles and suggested companies are aware of growing public skepticism about AI, particularly regarding job concerns. It raised concerns that influential research on AI's impact on work might increasingly come from companies with a vested interest in favorable conclusions.
Acemoglu also pointed to the lack of user-friendly AI applications comparable to earlier transformative software like PowerPoint or Word. He said that while people can chat with AI models, it often takes time for the average worker to achieve practical and productive use, contributing to AI's limited impact on the job market and economy thus far. He acknowledged significant uncertainty in the AI economy, contrasting it with the certainty of current rhetoric.
Full Article: Three things in AI to watch, according to a Nobel-winning economist