In an essay for *The Humanist Review of AI*, MIT professor Daron Acemoglu argues that AI should be developed to help workers perform their jobs better rather than primarily to replace them. He writes that current systems lack the reliability and versatility needed to automate whole jobs, and that adoption requires organizations to reorganize work and reallocate tasks. He also points to US tax incentives that, in his account, favor capital over labor and thereby subsidize automation. Acemoglu calls on developers and engineers to prioritize tools that complement workers. This is an argument about the direction and incentives of AI development, not a report of a new product or a measured outcome from a particular deployment.
Raw SignalOct 12, 20261 min read1 source
The case for a sidekick.
In one minute
# Daron Acemoglu argues for pro-worker AI over replacement