Predictive Analytics, 51 Loy. U. Chi. L.J. 161 (2019)
“Predictive Analytics” blends the latest research in behavioral economics with artificial intelligence to address one of the most important legal questions at the heart of intellectual property law and antitrust law - how do courts and agencies make judgments about innovation and competition policies? How can they better predict the consequences of intervention or non-intervention? The premise of this Article is that we should not continue to build doctrine at the IP-antitrust on theoretical neoclassical assumptions alone but also on the reality of markets using all that AI has to offer us. Behavioral economics and AI do not replace traditional antitrust analysis. Rather, they are complements and imbue antitrust law with continuing durability. Predicting competitive effects is difficult and we need tools to predict outcomes as precisely and reliably as possible. Until now, antitrust law has only been able to operate before a veil of assumptions and rhetoric. Stakeholders have only been able to think about whether and how to intervene in the exercise of IP rights, particularly patent rights, in the broadest terms since even the smallest perturbations in a complicated set of variables can set off ripples that lead to dramatically divergent outcomes. Facts have always mattered in antitrust law, and a more expansive toolkit can only increase our likelihood of getting it right. Behavioral economics sheds light on anticompetitive conduct that neoclassical antitrust may regard as irrational and therefore improbable. Once we recognize that it is rational and probable, we need to quantify and value the effects of the conduct. To do this, we need to employ more of the analogical reasoning intrinsic in antitrust law. For that, predictive analytics is very good in helping stakeholders with pattern recognition and simulation runs. This brings us closer to being able to ascribe value which human judgment can be brought to bear. In these, AI provides stakeholders with augmented capabilities to confront the computational challenges these tasks require.
Daryl Lim, Predictive Analytics, 51 Loy. U. Chi. L.J. 161 (2019)