What’s the right path for AI?

At an MIT symposium, journalist Karen Hao and scholar Paola Ricaurte delivered keynotes questioning the current trajectory of AI development. Hao argued that the massive scale-up of data, compute, and large language models is unnecessary and harmful, pointing to environmental costs, energy consumption, and the human toll on gig workers. She proposed an alternative vision centered on small, task-specific AI models like AlphaFold that tackle well-scoped problems with curated datasets. Ricaurte emphasized purpose-driven AI that responds to the real needs of communities. Both speakers called on the public to actively participate in shaping AI's direction.