The legal technology startup Fortuna Arbitration released its private court system with an artificial intelligence judge, called Arbitrus, this week after working on the tool for over a year.
In a paper published in the Social Science Research Network in January, the developers claimed their tool can cut down the average cost of dispute resolution from $100,000 to a flat $10,000, and lead to a system that they believe "will gradually outcompete" traditional dispute resolution methods. The Social Science Research Network is owned by Elsevier, a division of RELX Group, which also owns Law360.
As a result of their research, the developers claim that their AI tool did not produce any hallucinations, a common problem with other AI tools where the answer is misleading or false. The developers say Arbitrus breaks down evidence systematically and uses multiple models to determine the materiality, causal relationships, and temporal relevance of facts within a case. It also uses agentic AI, which involves using AI agents to make decisions.
The team behind Fortuna Arbitration includes Kimo Gandall, the CEO, and Kenny McLaren, the chief technology officer. Brian H. Potts, a partner at Husch Blackwell, is a co-founder.
While a tool that speeds up arbitration and is more cost-effective sounds good, questions remain about how the tool functions and whether the broader legal world will accept it.
Here, Gandall and Potts discuss how Arbitrus ensures accuracy and fairness, how they plan to convince parties to trust the tool, and the future of dispute resolution.
This interview has been edited for length and clarity
What ethical considerations guided your team's development of Arbitrus?
As we see it, our current public court system and arbitration from the American Arbitration Association are not adequate dispute resolution mechanisms for many claims. This is especially true for lower-dollar claims, where the cost of AAA arbitration or going to court outweighs the total value of the claim.
Our personal view is that Arbitrus could be the best thing that's happened to this country's dispute resolution system since arbitration was originally invented. Imagine how much more efficient and effective human decision-making could be if the courts and AAA were less clogged up with cases?
How does Arbitrus ensure the accuracy and fairness of its AI-driven arbitration decision, including on complex legal matters that require more nuanced judgment?
We use an iterative process where we carefully break down factual questions into conditions, similar to a decision tree. From there it gets passed into a voting ensemble and then into Ra.ai, Arbitrus' agentic AI solution, for an iterative justification.
On questions of law, Ra.ai is designed to optimize legal accuracy by mitigating hallucinations and enhancing decision-making through iterative validation. It initializes by determining jurisdictional relevance, employing vector indexing and cosine similarity to retrieve and rank case law. Multiple AI models cross-validate retrieved sources, ensuring proper legal grounding and eliminating erroneous citations.
We would note that not all arbitrators and/or judges are perfect. Far from it.
Arbitrus could issue 100 decisions in cases now, and if the same 100 cases went through human arbitration, Arbitrus would have a much higher percentage of accurate decisions based on existing law.
Do you expect challenges in convincing parties with high-value disputes to trust an AI arbitrator over a human?
There will be some challenges and some skepticism at first. But there are also likely a lot of litigations where both sides just want an answer quickly, and they could use Arbitrus now.
Once Arbitrus issues a bunch of rulings, it should be much easier to convince litigants that it's better than AAA arbitration or going to court.
We expect challenges, as all revolutionary technology does. But we are also introducing — in the next week or so — a prediction system that will allow parties to break down contract clauses, computer-simulate disputes and discern their probability of litigation ahead of time. This helps businesses determine the liability behind any contract or action.
Effectively, what an AI judge empowers us to do is industrialize the law. We are doing to arbitrators and lawyers what the Industrial Revolution did to blacksmiths. While human lawyers aren't going away, their trade will fundamentally change.
The best lawyers will continue to do what they have always done: advise their clients on the trends, risks, and opportunities of law — except now it will be quantitative.
What feedback have you received from early adopters of Arbitrus, and how has it influenced your product development?
We've spoken with at least a dozen general counsels interested in deploying Arbitrus into their vendor and/or employment agreements. We also have multiple advisers who helped shape the Arbitrus user experience and user interface.
How do you see the role of Arbitrus evolving alongside public courts as the future of dispute resolution takes shape over the next decade?
[Potts:] I envision a world that's not too far away where litigants only go to public courts if they must and/or if the claims are worth so much that they feel the need to involve a human judge.
Arbitrus.ai is to the law like the railroad was to the Industrial Revolution. It is the beginning of a radical social transformation in the means of governing society, distributing resources and ordering ourselves.
--Editing by Robert Rudinger.
For a reprint of this article, please contact reprints@law360.com.
