Brookings scholars lay out a people-first agenda for AI work

The most useful thing about a new Brookings essay on work and AI is that it flips the usual framing. The threat to workers, its authors argue, did not start with generative AI, and technology alone will not fix it. Job quality was already eroding, through what the Brookings piece calls 'increased monitoring and surveillance, algorithmic scheduling, and declining autonomy,' and the authors want the policy conversation to start there.

Their diagnosis borrows a piece of internet slang and lands harder for it: many workers 'already feel that their work is being degraded or, to use the language of the day, enshittified.' The piece pairs that with a familiar-feeling statistic, that more than half of surveyed Americans fear AI will take their jobs and replace their face-to-face relationships, but the deeper claim is that this fear is downstream of decades of weakened labor institutions, not of any specific model release.

From there the essay proposes three interventions. The first is minimum staffing rules, framed around education: 'laws that mandate minimum staffing levels and allocate funding to train the school workforce could bring more teachers into the classroom.' The second is a mid-career training system, on the argument that U.S. training 'has typically occurred either in schools, prior to employment, or on-the-job; neither approach works for mid-career workers in transition.' The third is tripartite institutions where government, business, and labor unions co-design how AI enters a workplace, with a small existing example the authors cite: computer scientists at Carnegie Mellon University and the UNITE HERE union co-designed an app for hotel guest room attendants that helps with communication and record-keeping.

The honest caveat is that this is an agenda, not a costed plan. The essay does not say how staffing mandates get funded, which agency would convene the tripartite bodies, or how retraining closes the pay gap between tech work and care work. It also leans on one classroom finding, that students using AI in class 'felt less connected to their teachers and peers,' that is worth taking as directional rather than settled.

What is useful here for anyone running a team or a policy shop is the reframing itself. If the interesting question is not 'how fast do we deploy AI' but 'which institutions do we rebuild so AI lands well,' then the near-term work is unglamorous: labor boards, staffing law, and design tables where the people doing the job get a seat before the pilot ships.