Why Society Needs a Proactive AI Strategy

In an article in the Spring 2021 issue of AI Magazine, Kochan makes the case that the effects AI will have on society “will depend on who participates in the key decisions influencing its design and use.”  Kochan is part of the faculty of the MIT Institute for Work and Employment Research (IWER) and was a member of the MIT Task Force on the Work of the Future.

In his article titled ”Artificial Intelligence and the Future of Work: A Proactive Strategy,” Kochan argues that AI vendors, left to their own devices, will more often develop the technology in ways that replace rather than assist workers. He predicts that in that scenario, AI will increase societal inequality and exact high social costs—and will be opposed by many workers.

Kochan also points out that prior research has found that technology strategies that are developed without engagement with the work systems in which they will be used tend to be less productive than those that are created with a more integrated strategy that engages end users in the technology’s development and implementation. Because of this dynamic, Kochan predicts that, if the trajectory of AI development is left strictly to technology vendors, AI is likely to underperform its potential in improving productivity.

But there is another path forward, Kochan argues.  To develop AI in a way conducive to shared prosperity, Kochan thinks that the views of multiple stakeholders need to be considered. “Much of the basic research that AI system developers build on is, or has been, funded with public resources,” Kochan writes. “Thus the public has a legitimate right to insist that AI systems address the public’s interest as well as the private economic interests of those purchasing or selling AI tools.”

Kochan notes that in a number of countries, including the European Union, are enlisting input from multiple stakeholder groups as they seek to establish AI governance guidelines.

U.S. labor law, Kochan writes, currently does not guarantee workers the right to negotiate over the design of technology before it is introduced in the workplace—something  he thinks need to be changed. And, he notes, worker representatives also need training so that they can participate effectively in decisions about new technologies in the workplace. To that end, Kochan is currently working on a project to develop an online class to teach workers and worker representatives about technologies such as AI.

By allowing workers and the public more input over the technological trajectory of AI, Kochan thinks many of the potential problems could be mitigated. “A new governance system could help AI achieve its full innovative and productivity-enhancing potential while ensuring that the benefits it generates are widely shared,” he writes.