Actually, Jack Dorsey, AI Doesn’t Have to Be This Way

Jack Dorsey's decision to lay off nearly half of Block's employees, citing AI efficiency gains, highlights a broader debate about the direction of artificial intelligence development. A working paper by economists Daron Acemoglu, David Autor, and Simon Johnson, titled "Building Pro-Worker Artificial Intelligence," critiques the current trajectory, arguing that most AI being built automates work or augments capital rather than extending worker capabilities. The authors identify five categories of technological change, concluding that only AI creating entirely new tasks is unambiguously good for workers; the others tend to favor capital owners under the current economic structure.

The paper attributes this trend to market failures, where companies are incentivized to cut labor costs through automation, developers are rewarded for creating interchangeable workers, and a cultural "pro-automation ideology" dominates the tech industry. This results in AI being designed against the interests of many workers. To counteract this, the authors propose nine policy directions, including public investment in worker-extending applications, tax reform to stop subsidizing automation over employment, antitrust enforcement, and crucially, intellectual property protections for worker expertise so that the knowledge used to train AI is not simply extracted to eliminate the jobs that generated it.

(Source：Hardresetmedia)