At a 2026 MIT conference, journalist Karen Hao critiqued the prevailing AI development trend toward massive data usage and model scale aimed at artificial general intelligence. She highlighted the significant environmental, social, and operational costs of such large-scale efforts, including energy consumption, data center emissions, and labor exploitation in data preparation.
Hao proposed an alternate path exemplified by targeted AI tools like AlphaFold, which deliver impactful results without extensive scaling. This task-specific AI model encourages more sustainable and broadly beneficial development strategies, moving away from unchecked data and compute expansion.
Source & further reading
What’s the right path for AI? ↗Source published 2026-03-20. This report was prepared with AI assistance from the linked primary-source material. Claims and limitations should be checked against the original announcement or research.