Artificial Neural Networks have reached “grandmaster” and even “super-human” performance across a variety of games, from those involving perfect information, such as Go, to those involving imperfect information, such as “Starcraft”. Such technological developments from artificial intelligence (AI) labs have ushered concomitant applications across the world of business, where an “AI” brand-tag is quickly becoming ubiquitous. A corollary of such widespread commercial deployment is that when AI gets things wrong—an autonomous vehicle crashes, a chatbot exhibits “racist” behavior, automated credit-scoring processes (...) “discriminate” on gender, etc.—there are often significant financial, legal, and brand consequences, and the incident becomes major news. As Judea Pearl sees it, the underlying reason for such mistakes is that “... all the impressive achievements of deep learning amount to just curve fitting.” The key, as Pearl suggests, is to replace “reasoning by association” with “causal reasoning” —the ability to infer causes from observed phenomena. It is a point that was echoed by Gary Marcus and Ernest Davis in a recent piece for theNew York Times: “we need to stop building computer systems that merely get better and better at detecting statistical patterns in data sets—often using an approach known as ‘Deep Learning’—and start building computer systems that from the moment of their assembly innately grasp three basic concepts: time, space, and causality.” In this paper, foregrounding what in 1949 Gilbert Ryle termed “a category mistake”, I will offer an alternative explanation for AI errors; it is not so much that AI machinery cannot “grasp” causality, but that AI machinery (qua computation) cannot understand anything at all. (shrink)

Economists and legal scholars routinely posit an implicit contract between Japanese firms and their principal lender. Under this arrangement, the bank implicitly agrees to rescue the firm when times turn bad. Out of court, it rescues the firm from insolvency. Not only does it save the investments specific to the troubled firm, it lowers the use of costly bankruptcy proceedings and cuts the costs of those bankruptcy procedures firms do occasionally invoke. Given the creditor-shareholder conflicts of interest that arise as (...) firms approach insolvency, such arrangements would seem unstable. Yet according to a long sociological tradition, conflicts of interest matter less in Japan than in the West. According to the emerging economic and legal tradition, Japanese economic actors do face those conflicts, but keep them in check through reputational concerns, close-knit ties, and government supervision. Using two datasets of troubled firms from the 1970s and 1980s, we ask whether Japanese main banks in fact rescue distressed borrowers. We find no evidence that they do: large Japanese firms fail; when large firms approach insolvency, main banks do not increase the share of the firm’s debt they bear; stronger ties between distressed firms and their main bank do not facilitate loans; and troubled firms do not try to preserve their main bank relationship. All told, the claim that Japanese banks ever implicitly agreed to rescue firms is sheer myth. That Japanese banks let troubled firms fail is no recent development; it has been thus for decades. Conflicts of interest do indeed matter in Japan and long have. They matter enough to prevent precisely the incentive-incompatible rescue deals that scholars in the field so routinely posit. (shrink)

On March 11, 2011, a magnitude 9.0 earthquake and thirty-eightmeter high tsunami destroyed Tokyo Electric’s Fukushima nuclear power complex. The disaster was not a high-damage, low-probability event. It was a high-damage, high-probability event. Massive earthquakes and tsunamis assault the coast every century. Tokyo Electric built its reactors as it did because it would not pay the full cost of a meltdown anyway. Given the limited liability at the heart of corporate law, it could externalize the cost of running reactors. In (...) most industries, firms rarely risk tort damages so enormous they cannot pay them. In nuclear power, “unpayable” potential liability is routine. Privately owned companies bear the costs of an accident only up to the fire-sale value of their net assets. Beyond that, they pay nothing — and the damages from a nuclear disaster easily soar past that point. Government ownership could eliminate this moral hazard — but it would replace it with problems of its own. Unfortunately, the electoral dynamics in wealthy modern democracies combine to replicate nearly perfectly the moral hazard inherent in private ownership. Private firms will build reactors on fault lines — but so will governments. (shrink)