By Leslie King O’Neal
Evolving AI Tools Complicate Law & ADR Practice
How can lawyers and mediators use AI tools wisely in mediation? During the last few years, the legal profession and the ADR world have struggled with using AI in practice. Is AI a wonderful tool or is it the death knell for lawyers, arbitrators and mediators? As AI tools emerge and evolve, their uses and abuses likewise change. At present no one can answer all the questions AI use legal practice and ADR practice raise. This post offers practical insights from an experienced ADR professional and AI user about using AI tools wisely in mediation.
See prior posts:
Practical Advice About Mediation AI Use from a Trustworthy Source
John Lande,[i] professor emeritus at the University of Missouri Law School and prolific author, was an early AI adopter and has written extensively on his experiences using AI tools in ADR. The latest edition of the blog Indisputably,[ii] features Professor Lande’s paper on Tips for Using AI Wisely in Mediation.[iii] Lawyers, mediators and parties to disputes can apply these tips when considering AI use to assist in preparing for mediation (or other tasks). Ask AI to help you think clearly, not to give you “the right answer” or to think for you
Guidelines For Effective AI Use in Mediation:[iv]
Provide all relevant facts including information that may not support your view.
- Don’t assume AI responses are correct – even if they sound confident.
- Ask AI to provide the materials it used so you can check it.
- Double-check legal information before relying on it.
- Don’t use materials created by AI unless you have checked them carefully.
- Treat what AI tells you as a starting point, not the final answer. Ask follow-up
- questions to consider other possibilities.
Think About the Case & the Circumstances
Ask AI tools questions, but make judgments based on the client’s values and goals. Don’t rely on AI tools to predict outcomes; instead ask questions like these:
- What does the client want to get out of this dispute (such as money, an apology, or a practical solution?
- How does the other side see this dispute?
- What information and documents should the client collect to prepare?
- What are the strengths and weaknesses of the case if it went to court?
- What agreements might both parties find acceptable?
- What would happen if the parties don’t reach agreement?
- What will it cost the client to continue (for example, in time, money, and stress)?
- If the case doesn’t settle, how might this affect the client’s relationships, peace of mind, financial situation, and other important things?
Don’t stop after the first answer. Ask follow-up questions
Recognize AI’s Limitations
By now most lawyers and mediators have read numerous horror stories about AI hallucinations. Any AI user should be aware that AI has serious limitations, especially in legal disputes. It can’t accurately predict what will actually happen in a case. Sometimes it makes up information, misunderstands questions, or gives information that is out of date. AI may not have complete or accurate information about the specific situation. For example, it may:
- Refer to cases or laws that don’t apply to the situation.
- Leave out important facts.
- Make assumptions that aren’t true.
- Sound like it’s always right – even when it’s wrong.
- Agree with you too easily instead of pointing out problems or other possibilities.
Always Verify AI Output
Before utilizing AI-generated content, be sure to check:
- Information that could affect an important decision.
- Court cases that AI mentions.
- Laws and legal procedures.
- Whether AI is using the correct law for your state or country.
- Court deadlines.
- Estimates of legal and other expenses.
Confirm answers that seem surprising or especially important, with a reliable source
Common Ways AI Is Misused
AI can’t answer every question about a case. Common ways AI is misused are:
- Asking AI What the Case Is “Worth”
Don’t ask AI to estimate how much money a party might receive or other possible outcomes. AI can’t realistically evaluate evidence, credibility, or legal standards accurately, and it may produce misleading numbers.
- Relying on AI-generated Legal Citations Without Checking Them
AI may generate court cases or statutes that are irrelevant, outdated, or completely incorrect. Some may not exist at all.
- Using AI to Decide What Offer to Make or Accept
AI does not understand the client’s risks, costs, or personal priorities. It cannot (and should not) make decisions about offers or acceptance.
Confidentiality Risks
Assume that anything you enter into an AI system may be stored or reviewed. If you are unsure, do not include identifying details. Do not enter:
- Real names.
- Addresses.
- Phone numbers.
- Government identification numbers.
- Financial account information.
- Medical information.
- Passwords.
- Confidential information learned during mediation.
When possible, use general descriptions instead of real names or other identifying information. Take a moment to think before clicking “send.” This can help protect everyone’s privacy. Check whether the AI service allows limiting or turning off the use of information for “training” or “improving” its system. This may reduce the risk that personal information will be used, but it may not eliminate that risk completely.
AI Chats and Prompts May Be Discoverable
Some recent cases have required AI chats and AI prompts to be produced in discovery and allowed them to be used as evidence. Do not assume that chats or prompts are private just because they were used to prepare for mediation.
See recent post:
Takeaways:
- AI can be a very helpful tool, but it is easily misused.
- AI can provide information, suggest ideas and guide thinking about options.
- Good decisions come from careful thinking about the situation. Verify AI responses.
- Use AI as a tool to assist in decision-making; don’t let AI make decisions.
For more information, see the Real Practice Systems Negotiation and Mediation Coach (RPS Coach) AI tool. It is a public service available for free.
[i] John Lande is the Isidor Loeb Professor Emeritus at the University of Missouri School of Law. He previously directed its LLM Program in Dispute Resolution. https://lande.missouri.edu/
[ii] Indisputably is a blog devoted to linking dispute resolution scholarship, education, and practice. It is supported by the Cardozo Law School and the Kukin Program for Conflict Resolution http://indisputably.org/
[iii]John Lande, Using AI Wisely in Mediation, Legal Studies Research Paper Series, Research Paper 2026:50, University of Missouri School of Law (June 29, 2026) https://papers.ssrn.com/sol3/papers.cfm?abstract_id=7019600 [1] The following is adapted from John Lande, Using AI Wisely in Mediation, Legal Studies Research Paper Series, Research Paper 2026:50, University of Missouri School of Law (June 29, 2026) https://papers.ssrn.com/sol3/papers.cfm?abstract_id=7019600
[iv] The following is adapted from John Lande, Using AI Wisely in Mediation, Legal Studies Research Paper Series, Research Paper 2026:50, University of Missouri School of Law (June 29, 2026) https://papers.ssrn.com/sol3/papers.cfm?abstract_id=7019600
Leave a Reply