AI in mediation: The modern Prometheus?

"Frankenstein" is all the rage. Everywhere one turns, plays, movies and media presentations purport to describe the famous monster. But for nearly 100 years, popular culture has misrepresented Mary Shelley's creation and her message.

Victor Frankenstein was fascinated not only by the physical and natural sciences but also by alchemy, the interplay of chemical and natural scientific inquiry with philosophical traditions. Over a two-year period, Frankenstein's endeavors led him to the dissecting rooms and slaughterhouses from which he gathered material to construct a humanoid being, one feature at a time, and eventually to animate it.

Frankenstein's creature learned to read, write, think independently and communicate like a human. It felt emotions, suffering the sting of rejection and isolation when Frankenstein and others spurned it as a grotesque devil. Despite its benevolent instincts, the creature was thwarted in its quest for companionship and affection. Others' revulsion ultimately led it to kill, destroying its creator and those whom he loved.

Mary Shelley wrote "Frankenstein; or, The Modern Prometheus" in 1818. Unlike sensational iterations of her work, the novel remains a morality tale about the promise and danger of scientific experimentation. It is a story of ambition, discovery, responsibility and unforeseen consequences.

Legal AI: Caveat emptor

"I had desired it with an ardour that far exceeded moderation; but now that I had finished, the beauty of the dream vanished, and breathless horror and disgust filled my heart." Victor Frankenstein, "Frankenstein" (Chapter 5).

We live in an era of rapid scientific and technological advancement. Conversation -- at conferences, in professional settings and in our daily lives -- is dominated by "artificial intelligence" (AI). General AI and artificial superintelligence could completely transform the way we live and work, but many fear the consequences of unregulated use.

Lawyers have already seen what can happen when they rely on AI without carefully monitoring its implementation: Attorneys have been sanctioned for citing fabricated authority or misrepresenting the law. But with billing rates escalating, legal clients are increasingly pushing use of AI in the name of efficiency.

As happened with the introduction of remote platforms, mediators and counsel are becoming increasingly less resistant to AI as they are exposed to its potential. Counsel now routinely use AI tools to digest and synthesize large quantities of information in litigation.

Databases have long been used to sort and organize. Humans coded and tagged topics, flagged privilege issues, and otherwise digested and analyzed documents before they could be used effectively. New tools permit comprehensive review without people having to make judgments about relevance and categorization. Increasingly sophisticated inquiries using AI tools allow content to be located and synthesized in a fraction of the time previously required.

AI in mediation: A new tool

"I was benevolent and good; misery made me a fiend. Make me happy, and I shall again be virtuous."  The creature to Victor, "Frankenstein" (Chapter 10).

The mediation community is now thoughtfully considering useful applications of AI. Five years ago, a global pandemic locked offices and prevented travel. Although most had never used online tools, mediators and counsel quickly adapted to the use of such platforms, learning the technology and its unique characteristics. Within a year, professionals who would have traveled to resolve cases were mediating remotely.

My first remote intellectual property mediation involved parties in China and the United Kingdom, and counsel in Los Angeles, Silicon Valley and Texas. The participants were not only in different time zones -- they were operating on different days. The matter settled, an agreement was drafted and the parties signed remotely. Today, a large percentage of mediations, especially those involving employment and personal injury disputes, are conducted remotely, and parties are spared uncomfortable encounters with one another in hallways or lunchrooms. 

Every document submitted to a mediator is confidential. Even when documents are exchanged among parties, no mediator would responsibly load those documents or exhibits into publicly accessible AI models that could use them to "learn" for other inquiries. Private mediation firms, which have closed systems, license AI software that facilitates mediators submitting queries, creating timelines, preparing summaries and posing questions that will prompt useful dialog to explore stakeholder interests, possible obstacles and settlement options.

As AI matures, it may become more reliable in analyzing submissions to probe strengths and weaknesses of the parties' positions and offer risk assessments. Counsel may then use similar tools to analyze positions and prepare for a mediation. Some AI platforms even purport to distinguish the motivation of the parties from nuanced language choices, revealing opportunities for integrative bargaining and creative solutions.

AI in mediation: Promise and peril

"Learn from me, if not by my precepts, at least by my example, how dangerous is the acquirement of knowledge." Victor Frankenstein, "Frankenstein" (Chapter 4).

Mediators spend countless hours -- usually without charge -- preparing for sessions. They review submitted material, read cases and statutes, analyze damage models, and prepare questions and issues for discussion. Before I mediate a case, I meet virtually with counsel on both sides to clarify and surface issues. Instead of creating timelines by hand while reviewing briefs, as I once did, I have a first draft generated by AI that I can edit for accuracy and comprehensiveness. AI can also generate a list of key names and relationships, a copy of essential contract terms and the text of statutes and regulations.

Models are being developed that will predict negotiation moves and outcomes. For simple distributive bargaining, these may prove useful, although parties will always need to test the resolve of their bargaining partners to see if a different result is possible from what experience suggested to counsel and the mediator early in the process. The historic "dance" of negotiation will never go away, but AI models show promise in predicting with accuracy the moves and results in many types of cases.

I regularly employ decision trees to evaluate potential settlement options. I may or may not share the details of these trees with participants, who tend to prefer a more high-level view, but as the number of branches and complexity of issues and outcomes proliferate, increasingly sophisticated methods of calculation are welcome.

A caveat to reliance on AI-driven models is that bias is inherent in all such systems. AI models are "trained" on data that must be rigorously overseen to minimize bias and ensure fairness. Users have noted inherent biases in screening employment applications, identifying preferential outcomes and other non-neutral orientations that a human mediator would strive to discern and resist.

AI in mediation: Less than human

If AI can synthesize documents and predict settlement options and negotiation outcomes, will mediators become obsolete? Data models are only as good as the information they process. As anyone who has attended a mediation knows, there will always be unanticipated turning points, most of which will be distinctly human.

One of my first mediations, 23 years ago, was a legal malpractice matter in which the defendant was self-represented. After hours of back and forth with defensive posturing and denial, I went into the defendant's caucus room unsure which tool to pull from my bag. After 30 seconds of silence, the attorney looked down, then back up and said, "OK." The subsequent confession led to a move that helped resolve the case.

Hours of trust building without judgment had created a safe environment for emotions to meet reality. AI may become better over time at reading facial expressions, gestures and behavior thought to reflect honesty. It may even increase its propensity toward creativity. But AI cannot be present at the right moment with the requisite sense of empathy and concern or the needed reality check. It is driven by its algorithms, not by the emotional intelligence that is the hallmark of effective mediators.

Conclusion

"Did I not, as his maker, owe him all the portion of happiness that it was in my power to bestow?"  Victor Frankenstein, Frankenstein (Chapter 17).

One hundred years ago, Mary Shelley dubbed Victor Frankenstein the "modern Prometheus." Prometheus was a Titan who created human beings out of mud and clay in the image of the gods. Zeus punished him for giving the precious gift of fire to humans by binding him to a rock, destined to have his liver plucked out every night by an eagle. Prometheus was ultimately saved from this eternal fate by a prophecy that instilled hope in the gods.

AI carries the twin burdens of anticipated promise and feared destruction. Mediators and counsel have the opportunity to integrate AI tools responsibly, recognizing the balance between innovation and compassion, and remaining the stewards of the tools they employ.