AI & Law Gary E. Marchant, J.D., Ph.D. gary.marchant@asu.edu What is AI? A machine that displays intelligent behavior, such as reasoning, learning and sensory processing. AI involves tasks that have historically been limited to humans and intelligent animals, such as decisionmaking and problem solving. 1
2
AI: Machine Learning Human Computer Programming Machine mechanically implements humanmade code Bad outcomes are attributable to bad code by human programmer Human programmer can explain why machine did what it did Machine Learning Humans provide data and specify overall goal for machine Machine self learns and adapts its approach to maximize specified goal Limited explanation for why machine did what it did 3
Including Law Artificial intelligence is changing the way lawyers think, the way they do business and the way they interact with clients. Artificial intelligence is more than legal technology. It is the next great hope that will revolutionize the legal profession. What makes artificial intelligence stand out is the potential for a paradigm shift in how legal work is done. AI and Law: Two Main Areas I. Substance of law - What law regulates II. Process of law - How law is practiced 4
I. Substance of Law Legal Regulation of AI Applications Superintelligence and existential risk Autonomous military weapons Autonomous cars Medical robots Algorithms Banking and financial services Discriminatory bots Criminal bots Sex robots 5
There s already an argument that being able to interrogate an AI system about how it reached its conclusions is a fundamental legal right. Starting in the summer of 2018, the European Union may require that companies be able to give users an explanation for decisions that automated systems reach. This might be impossible, even for systems that seem relatively simple on the surface, such as the apps and websites that use deep learning to serve ads or recommend songs. The computers that run those services have programmed themselves, and they have done it in ways we cannot understand. Even the engineers who build these apps cannot fully explain their behavior. 6
7
AI: Liability Possibilities Human Machine Civil Criminal Algorithm Concerns Criminal recidivism risk assessments Discrimination concerns Insurance Company algorithms? Discrimination concerns Search engine algorithms? Unfair competition concerns Data collection algorithms and Internet history algorithms? Privacy concerns; dynamic pricing Biometric algorithms? Bias and privacy concerns Securities and financial market algorithms? Market concerns; unintended collusion Banking/Lending algorithms? Market and discrimination concerns Service and product pricing algorithms (e.g., Uber) Consumer exploitation (e.g., surge pricing) 8
Two Bias Problems with Algorithms 1. Inaccurate bias algorithms may contain inaccurate biases against certain groups or individuals based on historical data that may be a reflection of historical bias e.g., historical data suggesting most successful engineers are white males 2. Accurate bias algorithms may contain empirically valid distinctions that we may not want to accept for fairness reasons e.g., young adult men are a higher car insurance risk than young adult women 9
Source: Pro Publica 10
11
Wisconsin v. Loomis Loomis was arrested for his role in drive by shooting, pleaded guilty to two charges involving unlawful driving In sentencing Loomis to jail rather than probation, trial judge relied (in part) on COMPAS risk assessment tool COMPAS uses a proprietary algorithm to predict risk of recidivism based on (1) age at first offense, (2) current age, and (3) criminal history Trial court, appellate court, and Wisconsin Supreme Court all rejected Loomis argument that reliance on proprietary algorithm violated due process Loomis currently ahs cert petition pending before US Supreme Court 12
13
14
AI Now 2017 Report: Privacy AI challenges current understandings of privacy and strains the laws and regulations we have in place to protect personal information. Established approaches to privacy have become less and less effective because they are focused on previous metaphors of computing, ones where adversaries were primarily human. AI systems intelligence, as such, depends on ingesting as much training data as possible. This primary objective is adverse to the goals of privacy. AI thus poses significant challenges to traditional efforts to minimize data collection and to reform government and industry surveillance practices. Significant shifts are needed in the legal and regulatory approaches to privacy if they are to keep pace with the emerging capacities of AI systems. II. Process of Law 15
16
Legal Research Can read over a million legal pages per minute Electronic Discovery According to academic studies, application of these methods (TAR) has resulted in significant monetary savings and time in locating ESI for disclosure in comparison to manual or human eyes review. Studies have shown that technology-assisted review is at least 50 times more efficient than human-review, or manual, review. Maura Grossman & Gordon McCormack, Efficient E- Discovery, ABA J. (April 2012). TECHNOLOGY- ASSISTED REVIEW (TAR) 17
Brief Writing Case Outcome Prediction 18
Jury Screening Billing and Staffing 19
Legal Marketing 20
Microsoft President and Chief Legal Officer, Brad Smith, added: If you can t afford a lawyer, then you can t solve crippling housing, child custody, or civil litigation disputes. Technology can help bridge this justice gap by empowering people with the advice and services they need to lead fruitful lives. Source: ArtificialLawyer.com 21
Replacing Courts?: On Line Dispute Resolution 22
23
24
25
So Who Will Win Humans or Machines? Human Lawyers? X 26
Machine Lawyers? X Correct Answer? + 27
Impacts on Young Attorneys The functions best done by AI will displace much of the work done by young associates at law firms e.g., document review, legal research Some large companies have already informed their law firms that they will no longer pay for the work billed by 1 st year associates Natural response of law firms will be to hire fewer entry level attorneys How will this affect professional pipeline and training for future lawyers? 28
AI and the Practice of Law Good News Adoption of AI in practice of law will be evolutionary, not revolutionary Most lawyer functions will not be performed by AI in foreseeable future Adoption of AI offers opportunity to be leader in efficiency, cost savings, accuracy Bad News AI will take over a steadily increasing share of law firm billable hours Disparate impact on young lawyers AI will require knowledge/abilities outside the existing skill set of most current practicing attorneys Incorporation of AI into practice will soon be a matter of keeping up rather than being a leader Advantage to Early Adopters? Unlike other game changers, such as mobile, where fast followers have benefitted from the lessons learned by early adopters, AI is one of those technologies where it pays to be a first mover because machine learning means that the system is continually improving as it learns from every matter and transaction. This means that early adopters systems will be better trained than those of their competitors. Joanna Goodman, Robots in Law (2016) 29
AI and Legal Malpractice Risks Attorneys will be responsible for AI system s mistakes, even though attorneys cannot fully understand or check what AI system does Will AI increase law firm liability risk? Most early adopters have reported that improved quality is bigger benefit than increased efficiency Will it be malpractice not to use AI? AI and New Legal Ethics Challenges Attorney A in Law Firm K trains and uses AI system on data from Client X Can client X ask firm K to share AI system with firm L representing X on another matter? Can law firm K use this AI system trained with client X with a new matter for client Y? If yes, how does firm K bill client Y? If attorney A moves to law firm L, can he take the trained AI system with him? Does it matter if she continues to represent client X? 30
Disruption of Law Firm Value Chain? If AI can do in a few seconds the work of an associate that now takes a couple weeks, how will this affect: Law firm billings? Continued use of billable hour? Partner/associate leverage? Will AI shift legal markets by favoring: Large firms over small/medium firms? In-house over law firms? Conclusion: How to Respond to Disruptive Technology? Clayton Christensen: Disruptive Innovation or 31