What AI Practitioners Should Know about the Law

recovery. Rules of tort liability mediate between the victim's need for recompense and the defendant's right to remain free from arbitrarily imposed obligation. Although sensational cases involving large recoveries tend to generate the greatest alarm, the magnitude of damages in a particular case is actually far less important than the availability of any damages in similar cases. Potential tort defendants are primarily interested in their overall liability exposure, as determined by the evolving structure of case precedent. For software developers, expenditures for debugging, design safety analysis, and quality control assurance are necessarily affected by the extent of perceived vulnerability to tort actions. More so than in any other legal field, the boundaries separating human from machine will be forced into focus by questions of tort liability. In addition to compensation, another function of tort law is to project proper standards of care in the conduct of activities that might cause harm. When the agent of injury is a tangible device or product, attention is currently directed toward three sets of possible culprits: manufacturers (Was the product designed or manufactured defectively?), sellers (Was the product sold in a defective condition?), and purchasers (Was the product used improperly?). Absent from the lineup is the injury-producing item itself, whose existence is relevant only insofar as it pertains to the conduct of human beings. As “devices” come to include electronic systems capable of judgment and behavior, they too will become objects of direct inquiry. Naturally, financial responsibility will ultimately rest with a lthough recent computer-related disputes have involved a variety of legal issues—breach of contract specifications, misrepresentation of performance capability, and outright fraud, to name a few—it is the threat of tort liability that continues to strike the greatest fear in the heart of mighty industrialist and struggling start-up alike. Once inside the courtroom, what role can the computer assume in its own defense or in the service of some other litigant? The law of evidence, developed to govern the testimony of human witnesses, must continually evolve to accommodate new, nonhuman sources of information. At the same time, developers of intelligent computer systems will have to achieve a great deal of communicative power before their machines will be given a day in court. These two topics—tort liability, and the computer as witness—are the focus of Part 2.