Joshua Graff Zivin is the Pacific Economic Cooperation Chair in International Economic Relations at UC San Diego, where he holds faculty positions in the School of Global Policy and Strategy and the Department of Economics. He is presently a research associate at the National Bureau of Economic Research and serves dual roles as director of the Cowhey Center on Global Transformation and co-director of the Global Health Institute at UC San Diego.
Dear Josh,
You have published multiple papers about innovation in science and about the design of incentives, as an economist and social scientist. Can you comment about the use of incentives to promote the sharing of ideas and collaboration in biomedical science?
Biomedical scientists, particularly those at universities and non-profit research institutes, face a number of incentives to advance science and share ideas. It is useful to split these into two types. Scientists are generally motivated by career concerns, which require outputs that can be attributed to an individual scientist or set of scientists to generate promotions and increase funding for their laboratories. Outputs of this sort include publications, citations, patents, as well as grants and prizes. They may also include a successful track record in hiring and placing students and postdoctoral researchers. At the same time, they are also often motivated by more altruistic concerns tied to the advancement of science and improved patient care, even when their contribution to those objectives cannot be easily observed or inferred by others.
Can recent developments in artificial intelligence offer new opportunities in this respect? Interactions with AI models contribute to their training and create value, which could be used to incentivize open discussions. At the same time AI will challenge the current way of doing things in biomedical research and it might be better for the scientific community to plan ahead to make sure these changes promote the common interest.
The promise of AI in the biomedical space is predicated on the availability of large-scale datasets that synthesize observations and findings from distinct laboratories that are unlike to otherwise see the linkages between their efforts. Apportioning credit to individual scientists will be especially difficult in this context and altruistic motivations alone will likely be insufficient to induce broad participation, particularly if these contributions crowd out activities and findings that might better align with their career incentives. To overcome this problem, we will need to find creative ways to allow scientists to extract more credit for these efforts. Scientific prizes could prove useful in this regard, with the usual caveats that prizes only work when the criterion for their award is well specified ex ante. In this case, however, there is a second challenge. Evenly sharing the prize amongst all contributors will induce some to underprovide effort – the well-known free-rider problem. A better system would perhaps provide all participants with a baseline reward and then offer a pro-rated bonus prize based on relative contributions. While such apportionment is difficult, AI and other data science tools may be able to help by identifying statistically the relative value of individual contributions that led to a given discovery. The exact features required to induce sufficient participation remains an open question and our earlier work suggests that this will depend greatly on the desired balance between incremental and more radical discoveries. Both are clearly important and financial incentives (as well as less pecuniary ones) can have a profound impact on the likelihood of generating either (Graff Zivin and Lyons 2021; Azoulay et al., 2011). Ultimately, this should be explored through experimentation by varying prize amounts and sharing rules across similar activities in a manner that allows one to infer the strength and limitations of each incentive scheme (Azoulay et al., 2013).
REFERENCES
Azoulay, P, J Graff Zivin, and G Manso, “Incentives and Creativity: Evidence from the Howard Hughes Medical Investigator Program” The RAND Journal of Economics, 42(2011): 527-554.
Azoulay, P, J Graff Zivin, and G Manso, “NIH Peer Review: Challenges and Avenues for Reform,” in Innovation Policy and the Economy, Volume 13, J Lerner and S Stern (Eds.), University of Chicago Press, 2013.
Graff Zivin J and E Lyons “The Effects of Prize Structures on Innovative Performance,” American Economic Review: Papers and Proceedings, 111(2021): 577–581.