AI workplace transformation will be slower and less complete than tech insiders predict
Source: https://www.nytimes.com/by/ross-douthat. "Opinion | How Fast Can A.I. Change the Workplace? - The New York Times." February 14, 2026. www.nytimes.com
The Gist
While AI will definitely change jobs, it won't happen as fast as tech experts think because society is complicated and people like dealing with real humans. History shows new technology creates new jobs rather than eliminating work entirely, though AI might be different because it acts more human-like.
Conclusion
While AI will significantly impact employment, the transformation will be slower than AI insiders predict and won't lead to mass unemployment due to social frictions and human preferences
Premises
- AI models are clearly capable of eventually replacing many human jobs across white-collar professions
- Human society creates complex bottlenecks that slow even the most efficient innovations through false starts, misadaptations, and institutional resistance
- Employment relationships involve complex contractual, social, legal and bureaucratic factors beyond pure productivity maximization
- Historical precedent shows that technological innovations have consistently created new forms of work rather than permanent mass unemployment
- People demonstrate clear preferences for human interaction even when automation is available, as seen with piano players, cashiers, and waiters
- AI's unique ability to simulate human-like interaction makes it fundamentally different from previous technologies
- The extent to which people relate to AI as conscious beings will determine how willing they are to accept AI replacement of human workers
Assumptions
- Historical patterns of technological adaptation will continue to apply to AI
- Social and institutional frictions will remain significant barriers to rapid change
- Human preferences for authentic human interaction are deeply rooted
- Companies will prioritize factors beyond pure efficiency when making employment decisions
- The pace of AI development can be predicted based on current trajectories