The history of AI has not been kind to pioneers who overpromise and underdeliver. The boom-bust cycle of AI has seen exciting and powerful new techniques developed during AI booms, only to be followed by an AI winter. In this article, Clive Thompson examines the potential for a new AI winter and the risks it poses to the field. He explores the current state of AI development and the factors that could lead to another winter, including unrealistic expectations, overselling the capabilities of AI, and a lack of investment in basic research. Thompson argues that the consequences of another AI winter could be severe, and that we need to take a greater level of care and caution in the development and implementation of AI. As AI continues to evolve, it will be up to us to steer it towards a sustainable and beneficial future.
