How Academic Labs Adapt to Advances in AI: The Impact of AlphaFold

Author: Gabriel Cavalli

Last updated on: September 2nd, 2024

Abstract:

This study investigates how principal investigators (PIs), acting as managers of their academic laboratories, adapt their organizations in response to an advance in AI that improves solutions to a scientific problem they were already addressing in computational biology. PIs differ in their specializations. Some are generalists with multidisciplinary academic backgrounds in both life and computer sciences, while others are specialists with expertise in one single discipline. This variation influences their decisions as organizational managers regarding size and expertise composition of their labs. In response to an advance in AI, generalist PIs are hypothesized to adapt by reducing lab size while balancing organizational expertise towards generalism. Specialist PIs are hypothesized to expand lab size and recruit additional computer science specialists. These hypotheses are tested within the setting of academic labs participating in the CASP competition (Critical Assessment of Protein Structure Prediction). A surprise occurred in 2018 when Google DeepMind’s entry, called AlphaFold, employed AI to achieve unprecedented success in modeling the “protein-folding” problem. By using a manually compiled dataset sourced from lab websites and LinkedIn, the findings indicate that PIs adapted as hypothesized based on their specializations. Notably, the analysis also reveals that PIs with specializations in the life sciences not only recruited more computer science but also more life science specialists. The results emphasize the crucial role of managerial judgment, informed by expertise, in shaping responses to technological innovations incorporating AI, and in integrating human and artificial intelligences within organizations more broadly.

Orchestrating Innovation in Pharmaceutical Science in Low- and Middle-Income Countries after TRIPS

Authors: Gabriel Cavalli, Michael Blomfield, Anita McGahan, Keyvan Vakili

Last updated on: June 26th, 2024

Treating disease in LMICs is a Grand Challenge in healthcare. To meet this challenge, many LMIC governments agreed to implement a harmonized patent system called TRIPS that creates incentives under a market-pull logic for pharmaceutical companies to develop and sell drugs locally for relevant diseases. Yet this outcome has not been fully achieved, leading to conjectures that institutions must first develop to support both the diagnoses that generate drug demand and the scientific practices that stimulate drug supply. Relying on abductive reasoning, we find evidence supporting both conjectures by analyzing how scientific activity changes in countries at different levels of development after TRIPS implementation. As public policy, TRIPS stimulates market developments that may eventually lead to effective treatments for diseases prevalent in LMICs.