Philipp Kircher

Professor of Industrial and Labor Relations

Philipp Kircher, is a professor of industrial and labor relations, and a founding member of Cornell University’s AI-Ready Workforce Initiative. This short Q&A with Kircher picks up some of the big un-answered questions related to AI and labor.

Q: Who gets the productivity gains from AI? 

“Some research suggests that the gains currently go to more experienced workers within professions, at the costs of less experienced workers. Newcomers into an occupations seem to have a harder time getting in (even though their productivity seems to improve fastest with the use of AI.  

Q: Does AI increase workers' bargaining power or employers? 

“It depends on the occupation you are in. For workers in high demand, such as AI specialists or those who maintain large computing centers it is likely that bargaining power has increased because their skills are scarce. Obviously for workers who are in jobs substituted by AI the opposite is the case. More interesting is the gig economy. Algorithms that use detailed information about workers could, in theory, set each worker’s pay at the lowest amount they would be willing to accept, weakening their bargaining power. A lawsuit against Google just got filed in the Netherlands alleging this. 

Q: Will AI make workers more productive or simply reduce demand for labor? 

“I think it is both. Workers become more productive and that replaces some demand for labor. It is noteworthy that the improvements in productivity observed in the aggregate are not as large as some have predicted yet. The reason might be that lots of firms use some AI, but few have thoroughly integrated it into the depth of their business practices.

Q: Could technology make shorter workweeks economically viable? 

“In principle yes, but it is not clear that workers would want this. Evidence from the US suggests that most workers would like to work more rather than less – arguably to increase their overall paycheck. So, reducing the workweek is only sensible if the productivity gains of AI translate into wages rather than into firm profits. (In Europe the situation is different – many workers there would like to work slightly less at constant hourly pay). 

Q: What happens to wages when AI substitutes for some tasks but complements others?

“That is arguably what is happening. This is fine as long as workers can shift in their current job into complementary tasks, and if demand for the final product increases enough to keep the labor employed. It becomes problematic if the complementary tasks are in other positions, as this will require substantial labor reallocation not just across tasks but also across jobs, occupations and potentially across employers. Helping workers to understand which situation they are in, what alternative skills and options they have, and how to make a transition, will be key for a successful transition into the AI age.”

JR Keller

Assistant Professor of Human Resource studies

JR Keller is a professor of human resource studies at Cornell’s School of Industrial and Labor Relations and a founding member of the AI-Ready Workforce Initiative. He says employees have more trust in employers who disclose when and how AI is used in the workplace.

Keller says:

“At this point in time, even a general acknowledgement of where AI is being used and where it is not would be welcomed by workers. Most organizations can’t even explain exactly how the AI works, so the expectation that they provide detailed information on what data is used and how seems a step too far. But candidates want to know, and firms should be able to give some information to applicants to avoid the best candidates opting out of applying simply because they assume AI is being used unfairly.  

“It is hard to imagine things like traditional job descriptions going away, or even undergoing a major evolution. I do, however, imagine that softer capabilities like decision-making and influencing will become more desirable and show up more frequently, as these are what are needed to work most effectively with AI and make informed decisions about where and when it makes the most sense to deploy AI.”