NYU Tandon professor helps devise a way to efficiently incorporate AI into the workplace
Anne Laure-Fayard, associate professor of technology management and innovation, is the co-author of a new report in the Harvard Business Review that takes a sharp look at challenges to implementing AI systems in the workforce. The report enumerates key methods of successfully achieving the task, which requires technologists and administrators to think deeply about the relationship — and potentially antagonistic reactions — employees may have to machine learning-based systems.
While the article, “A Better Way to Onboard AI,” co-written by INSEAD faculty Boris Babic, Daniel L. Chen, and Theodoros Evgeniou, cites a 2018 Workforce Institute survey of 3,000 managers across eight industrialized nations finding that most described artificial intelligence as a valuable productivity tool, it also notes that respondents harbored fears that AI would take their jobs; also noted is a recent report that showed that more than 6 million workers in the U.K. fear being replaced by machines.
The authors explain that introduction of AI into the workflow must be done in a way that, to some, extent mirrors how new (human) hires are trained on simpler tasks that allow them to build hands-on experience in a noncritical context before being paired with mentors. The goal, in the case of AI, is not only to train the machine but to acclimate the employees and familiarize them with the benefits of AI tools.
Drawing on research and insights from AI and information systems implementation, along with organizational studies of innovation and work practices, the authors devised a four-phase approach to implementing AI, one that allows enterprises to cultivate people’s trust — a key condition for adoption — and to work toward a distributed human-AI cognitive system in which people and AI both continually improve.
- The first phase of onboarding AI involves training on fundamental rules and assignment of basic, time-consuming tasks, such as sorting data
- The second step requires setting up AI systems to provide real-time feedback. AI can be trained to accurately forecast what a user’s decision would be in a given situation. Thus, if a user is about to make a choice that is inconsistent with his or her choice history, the system can flag the discrepancy and share it with the user to allow them to reflect on the decision
- The third phase casts AI in a coaching role, which the authors suggest could benefit employees through a careful analysis of key decisions and actions and an ability to question AI’s recommendations
Finally, the authors cite cognitive anthropologist Edwin Hutchin’s theory of distributed cognition to assert that full implementation of AI in a partnership role with employees is intrinsic to how humans create “extended minds” through relations with social groups and inanimate objects. The theory asserts that cognitive processing is not necessarily limited to the brain, or even the body: External tools and instruments can, under the right conditions, play a role in cognitive processing and create what is known as a coupled system. Ship navigation is an example offered by Hutchins, since, in bygone eras at least, it constituted a coupled system involving such navigation tools as sextants, charts, and astrolabes, and, of course, sailors.
While the authors argue that artificial intelligence may be perceived as particularly difficult to implement, if done mindfully, adoption can be fairly smooth.
“A Better Way to Onboard AI” can be found at hbr.org/2020/07/a-better-way-to-onboard-ai