Artificial Intelligence and Arbitration: A Perfect Fit?
By Robin Dodokin, Sarah McEachern, Les Honywill
The pandemic pushed the legal profession to consider our relationship with technology and embrace it in day-to-day practice. The rapid adoption of innovative technology leads to the question of whether artificial intelligence (AI) will make further inroads into alternate dispute resolution. Machine learning and AI have progressed so far that their integration into the arbitral process seems inevitable, with the only question being a matter of time and degree. For instance, will there come a time where parties are arguing their case in front of a series of algorithms represented solely by an avatar? Such a reality may be closer than we imagine.
The pandemic pushed us to consider our relationship with technology and embrace it in our work. The rapid adoption of innovative technology leads to the question of whether artificial intelligence (AI) will make further inroads into alternate dispute resolution.[1] Aspects of machine learning are already integrated in lawyers’ practices and ADR. Algorithms help lawyers research issues in a fraction of the time it would have taken in the library. Judges have started to incorporate AI to help with sentencing decisions. Legal technology start-ups claim the ability to predict case outcomes. Massive document review projects can be expedited significantly.
Machine learning and AI have progressed so far that their integration into the arbitral process seems inevitable, with the only question being a matter of time and degree. For instance, will there come a time where parties are arguing their case in front of a series of algorithms represented solely by an avatar? Such a reality may be closer than we imagine.
1. Use of AI in adjudication
AI has the potential to enhance dispute resolution processes. It is particularly adaptive to mediation and high-volume consumer disputes. For example, AI could help negotiating parties identify areas of overlapping interest, creating a faster and cheaper route to a mutually beneficial resolution. Ebay and Paypal resolve millions of disputes every year through algorithm-driven online dispute resolution (ODR). The ODR algorithms leverage vast amounts of historical dispute data to diagnose the problem, based on similar disputes, and suggest mutually beneficial outcomes. These ODR systems also continue to “learn” and improve through user feedback.[2]
Not only is AI being used to recommend outcomes to parties, but some jurisdictions are also using AI to make binding decisions. China’s Internet Courts feature AI judges that take evidence, ask questions, set schedules, and issue rulings.[3] Courts in Michigan, Ohio, California, Wisconsin and Utah have implemented algorithm-based ODR, primarily for small claims, traffic violations, outstanding warrant cases, and low-conflict family court cases.[4]
2. Benefits of AI
The use of AI in these settings demonstrates the enormous benefits such technology could bring to the arbitration field.
A welcome benefit of AI in adjudication would be the reduction of bias in selecting neutrals to the extent that the parties wish to achieve that goal. AI arbitrator selection from a database of arbitrators may help to eliminate bias.
Also, recently the Ontario Bar Association reported that in their survey Ontario commercial litigators selected male arbitrators 82.6% of the time and white arbitrators 100% of the time; and selected male mediators 80% of the time and white mediators 98% of the time .[5] Perhaps mediator and arbitrator selection via AI would ensure more diversity in the selection of neutrals.
AI technology could be useful in mediation in the lead up to arbitration by offering real-time assessments of settlement offers against the strength of each party’s case. However, negotiations often result in outside-of-the-box solutions that meet parties’ non-monetary interests. Can algorithms and AI properly assess creative solutions? Remember the orange dispute between two family members? Dividing the orange in half did not fully meet the parties’ interests as one wanted the juice and the other wanted the skin of the orange to make marmalade. Could AI get the parties to this nuanced solution?[6]
The technologies used by other adjudicative processes offer the potential for substantial improvements in arbitration efficiency, which is a primary reason why parties choose arbitration. AI technology could eventually eliminate human error, reduce cost, and introduce greater impartiality into the arbitration process.
We are unaware of any commercial arbitrations currently being conducted solely by AI. But AI is assisting arbitration parties to review and produce documents in eDiscovery, analyze data, see patterns, and summarise legislation and case law. Companies such as Lex Machina, Arbilex, Arbitrator Research Tool (ART), Lit-gate and Arbitrator Intelligence are integrating AI technology into legal research. Parties can use AI to select arbitrators, counsel, or experts. Technology helps parties decide on the best panel or best expert for their legal issue.[7] AI has long been used in the eDiscovery process to review and produce documents. If companies such as Blue J L&E are able to use their technology to predict legal outcomes at high rates of success, could an arbitration process use the same technology to render the decision?[8]
3. AI’s limitations
AI technology and arbitral users may not be ready for this leap just yet. Predictive technology often bases its assessments on the judgment patterns of specific adjudicators, and some jurisdictions are pushing back out of a concern that the data mining would raise privacy and confidentiality issues.[9] In addition, data mining of arbitration awards is thwarted because most awards are known only to the parties involved.
Another limitation is AI’s limited ability to handle the intangible, complex, and unpredictable elements of arbitration. An arbitrator has to decide on the issues of the case, which involves reasoning and recognition. Cognitive biases also play a role in the decision-making process. Decision-making is a human trait. Is it possible for AI processes to reason as required in an arbitral award? Technology can make predictions based on previously recognised patterns but can technology reason as humans do? Can technology consider novel legal questions that involve a complex factual matrix or legal tests? Can AI technology write a well-reasoned award that will satisfy arbitral users and be enforceable? We do not have the answers to these questions beyond “not yet.”
Confidence and trust in the decision maker and ADR process is a central tenet of an effective dispute resolution system. We trust that a competent arbitrator will listen to and consider the evidence and argument through a fair process and then apply the correct legal principles in a well-reasoned award. On the one hand, knowing that an artificial adjudicator would be less likely to suffer from human error, bias, or distraction, may be of some comfort to parties. But will parties trust a virtual arbitrator where its thought process is shrouded by a “black box”? Can parties ensure that the programmer’s biases have not seeped into the algorithm rendering the decisions?[10]
4. Thoughts for the future
The pandemic accelerated the adoption of technology in a near overnight transformation. The arbitration community, which holds efficient and cost-effective dispute resolution among its goals, should be open to embracing technology (including AI), that achieves these aims. Questions certainly remain about AI integration and whether technology can fully meet the demands and gain the trust of arbitration users. However, the promise presented by AI and the disruption caused by the pandemic make the increasingly rapid adoption of AI-powered legal technology seem inevitable. The implications of AI and its reliance on probabilities will have to be carefully considered by the legal profession. That said, certain functions such as non-consensual arbitrator selection, assessing risk or the probability of an event occurring, and recognising patterns in documents via technology may be welcome.
[1] “AI” does not have a single, widely accepted definition. Artificial intelligence generally refers to a broad spectrum of processes designed to teach human intelligence to machines, enabling those machines to imitate human decision-making processes. The World Intellectual Property Organisation (WIPO) defines artificial intelligence as “a discipline of computer science that is aimed at developing machines and systems that can carry out tasks considered to require human intelligence. Machine learning and deep learning are two subsets of AI. With the recent development of new neural network techniques and hardware, AI is usually perceived as a synonym for ‘deep supervised machine learning.’” See AI and Intellectual Property”, online: World Intellectual Property Organization [https://perma.cc/8V68-LK6T]; Jenny Gesley, “Artificial ‘Judges’? – Thoughts on AI in Arbitration Law” (13 January 2021), online: Law Librarians of Congress [https://perma.cc/HH5E-S72Q].
[2] Colin Rule, “Resolving Disputes in the World’s Largest Marketplace” (Fall 2008), online: Colin Rule [https://perma.cc/2C9B-2JRK]; Coglianese, Cary and Dor, Lavi M. Ben, “AI in Adjudication and Administration” (2021). Faculty Scholarship at Penn Law. 2118 at p. 812.
[3] Tara Vasdani, “Robot justice: China’s use of Internet courts”, online: LexisNexis [https://www.lexisnexis.ca/en-ca/ihc/2020-02/robot-justice-chinas-use-of-internet-courts.page]; Monisha Pillai, “China Now Has AI-Powered Judges”, online: RadiiChina [https://radiichina.com/china-now-has-ai-powered-robot-judges/]. For examples of AI implementation in adjudication in Colombia see: Irma Isabel Rivera, “The implementation of new technologies under Colombian law and incorporation of AI in judicial proceedings”, online: International Bar Association [https://www.ibanet.org/article/14AF564F-080C-4CA2-8DDB-7FA909E5C1F4].
[4] John Villasenor & Virginia Foggo, “Artificial Intelligence, Due Process and Criminal Sentencing” (2020) 2020:2 Mich St L Rev 295.; Coglianese supra note 3 at p. 812; A. D. (Dory) Reiling, “Courts and Artificial Intelligence” (2020) 11(2) International Journal for Court Administration 8 at p. 6-7.
[5] Report by the OBA Working Group on Neutral Diversity, “Neutral Diversity in Ontario”, available at https://www.oba.org/CMSPages/GetFile.aspx?guid=feddadcd-980f-4e83-86c9-ff8ecbfba32d.
[6] Lee Jay Berman, “The Orange Story”, (originally published in Santa Monica Business Journal in May 1996)[http://www.mediationtools.com/articles/smbj9605.html]
[7] Aditya Singh Chauhan, “Future of AI in Arbitration: The Fine Line Between Fiction and Reality” (26 September 2020), online: Kluwer Arbitration Blog [http://arbitrationblog.kluwerarbitration.com/2020/09/26/future-of-ai-in-arbitration-the-fine-line-between-fiction-and-reality/]
[8] For a survey of technology resources for arbitration practitioners see the International Bar Association’s report on its website available at: https://www.ibanet.org/technology-resources-for-arbitration-practitioners.
[9] Andy McDonnell & Stephen Traynor “France Bans Analytics of Judges’ Decisions” (21 June 2019), online: Lexology[https://www.lexology.com/library/detail.aspx?g=ff53dfbe-0fe6-4dee-8a1d-990bf8459020]
[10] Sophie Nappert, “The challenge of AI in arbitral decision-making” (4 October 2018), online: Practical Law UK[https://uk.practicallaw.thomsonreuters.com/w-016-8848?transitionType=Default&contextData=(sc.Default)&firstPage=true]
Robin Dodokin has many years of dispute resolution experience as a commercial litigator, mediator, and arbitrator based in Toronto Ontario. She is a Fellow of The Chartered Institute of Arbitrators (FCIArb.) and has been designated as a Qualified Arbitrator (Q.Arb.) and Qualified Mediator, (Q.Med.) by the ADR Institute of Canada. Robin is a roster arbitrator and mediator at Arbitration Place and also a co-editor of “Arbitration and Business Cases”, which summarizes important arbitration and business cases.
Sarah McEachern practices commercial arbitration as arbitrator and counsel at Borden Ladner Gervais LLP. She is a roster panelist at VanIAC and a member of the NextGen arbitration roster at Arbitration Place. She is a co-author of the IBA’s report on technology resources for arbitration practitioners.
Les Honywill is a commercial litigation lawyer at Borden Lardner Gervais LLP’s Vancouver office with a focus on arbitration, trusts, and tax disputes. He has a particular interest in the interaction of the law and technology and how this relationship changes the way that we apply and practice the law.