Get the latest tech news
Harvard, MIT, and Wharton research reveals pitfalls of relying on junior staff for AI training
New study by Harvard, MIT, Wharton, and BCG researchers finds that relying on junior employees to train seniors on generative AI risks is ineffective, highlighting the need for top-down governance and expertise at all levels.
The research, conducted by academics from Harvard Business School, MIT, Wharton, and other institutions in collaboration with Boston Consulting Group, found that junior employees who experimented with a generative AI system made recommendations for mitigating risks that ran counter to expert advice. But the study found the junior employees’ risk mitigation tactics were often grounded in “a lack of deep understanding of the emerging technology’s capabilities,” focused on changing human behavior rather than AI system design, and centered on project-level interventions rather than organization or industry-wide solutions. “To explain how and when junior professionals may fail to be a source of expertise in the use of an emerging technology for more senior members, we must take into account not only status threat, but also risks to valued outcomes,” the researchers wrote, noting AI’s exponential rate of change, superhuman capabilities, and reliance on vast amounts of data.
Or read this on Venture Beat