The authors, MIT economists Daron Acemoglu and Simon Johnson, argue that the current path of AI development is not conducive to shared prosperity or democracy. They believe that instead of trying to replace humans with AI, the focus should be on making machines useful to people and creating technologies that generate marginal productivity gains, leading to more job opportunities.
The authors point out that many AI programs are limited and cannot perform tasks that involve social interaction, adaptation, flexibility, and communication as effectively as humans. They argue for a shift towards a philosophy of “machine usefulness” in AI design, where tools are put in the hands of workers, including lower-wage workers, to enhance their capabilities and value in the workforce.
Acemoglu and Johnson also highlight concerns about the use of AI in surveillance, facial recognition, data collection, and the spread of misinformation. They argue that these AI applications can reinforce power imbalances and undermine democratic participation. To address these issues, the authors propose various policy responses, such as data ownership for internet users, tax reform that rewards employment over automation, diversified high-tech research support, and measures to limit the profitability of algorithm-driven misinformation.
Read more on the MIT News Website or purchase their new book Power and Progress