Here is the podcast on my new article that was just included on the EDRM Global Podcast Network. Echoes of AI: Episode 6 | Dario Amodei’s Essay on AI, ‘Machines of Loving Grace,’ Is Like a Breath of Fresh Air. Google’s Gemini AI writes and creates the podcast, not me. All I do if direct the AIs and verify that they got it right. This usually requires several takes and my hands-on direction of these temperamental AIs, but it is an interesting, new way to learn. These AI podcasters often provide insights on my articles that I missed.
Artificial Intelligence is no longer just a tool for automating mundane tasks—it’s now stepping into arenas traditionally dominated by human judgment and empathy. One of the most intriguing applications of AI is in dispute resolution, where large language models like GPT-4 are being tested as mediators. Pre-trial settlements are critical to the continued functioning of our system of justice because an estimated 92% percent of civil cases are resolved out of court. With the rise of online dispute resolution, the potential for AI to resolve low-stakes disputes autonomously is appealing, especially as legal systems are increasingly overburdened with new case filings.
AI image by Ralph Losey using WordPress’s Stable Diffusion
But can AI truly manage the complexities of human conflict? What happens when a machine has to balance neutrality with empathy, or data analysis with human emotion? This article discusses a groundbreaking study, “Robots in the Middle: Evaluating Large Language Models in Dispute Resolution,” offering insights into how AI may augment—though not replace—human mediators. Let’s explore the future of AI in the courtroom and beyond.
AI image by Ralph Losey using WordPress’s Stable Diffusion
Introduction
The “Robots in the Middle“ study provides an empirical evaluation of LLM AIs acting as mediators. Mediation is very important to our system of justice because mediation and other methods of voluntary settlement keep our court systems functioning. According to an article by Harvard Law Professors, David A. Hoffman and John H. Watson, Jr: “… up to 92 percent of cases are resolved out of court, a figure that does not include the number of lawsuits that are never filed because the parties used other dispute resolution methods at the outset.” Resolving conflict outside the courtroom: Why mediation skills are increasingly valuable for lawyers, according to two Harvard Law experts (Harvard Law Today, 4/29/24). This is one reason why AI researchers in legal technology are so interested in the possible application of LLM AI to mediation.
The “Robots in the Middle“ study was based on mediation by text messages of disputes in fifty hypothetical disputes. The analysis and responses of humans with AI expertise and some limited legal experience, none of whom were professional mediators, were compared with responses of ChatGPT4o (omni). The AI prompts used in the experiment were based on the mediator’s guide of the Department of Justice of Canada. Dispute Resolution Reference Guide: Practice Module 2(August 25, 2022). The humans and AI were asked to select among thirteen mediation types set forth by the DOJ Canada. They were asked to pick one to three types they judged to be appropriate for each hypothetical. Then they were asked to prepare text messages to facilitate a settlement. The authors served as blind judges to evaluate the quality of the responses, not knowing which were generated by humans and which by ChatGPT4o.
AI image by Ralph Losey using WordPress’s Stable Diffusion
With the growing demand for Online Dispute Resolution (‘ODR”) platforms (see e.g.ODR.com), the study examines whether LLMs like GPT-4 might be able to effectively intervene in disputes by selecting appropriate types of mediation interventions and drafting coherent, impartial intervention messages.
The premise is simple, yet potentially transformative: if AI can handle routine, low-stakes disputes efficiently, this would alleviate the burden on human mediators, allowing them to devote their time and expertise to more complex, emotionally charged cases. The research sought to answer three fundamental questions:
How well can LLMs select intervention types in disputes?
How do LLMs compare to human mediators in drafting intervention messages?
Are AI-generated messages safe, free from hallucinations, and contextually appropriate?
AI image by Ralph Losey using WordPress’s Stable Diffusion
Who is Behind this Study?
The “Robots in the Middle“ study has seven authors in academic fields of law and technology: Jinzhe TAN, Hannes WESTERMANN, Nikhil Reddy POTTANIGARI, Jaromır SAVELKA, Sebastien MEEUS, Mia GODET and Karim BENYEKHLEF. They are from multiple Universities: Cyberjustice Laboratory, University of Montreal, Canada; Maastricht Law and Tech Lab, Maastricht University, Netherlands; Mila – Quebec AI Institute, University of Montreal, Canada; School of Computer Science, Carnegie Mellon University, United States; and,Faculty of Law and Criminology, Universite Libre de Bruxelles, Belgium.
These are all brilliant international scholars with great expertise in legal theory and AI technology, but none appear to have any actual experience as a mediator or even experience serving as an attorney advocate in a mediation. Only one of the authors appears to have any experience with the U.S. legal system, Jaromır Savelka, who is a researcher associate at Carnegie Mellon. Savelka previously worked as a data scientist for Reed Smith (2017-2020). The lack of real legal experience in dispute resolution by the human subjects in the experiments is a weakness of this study.
Since my article is written primarily for U.S. attorneys and legal tech experts, I try to correct for this gap with input from a certified mediator, Lawrence Kolin, who has mediated thousands of cases of all types since 2001. He is also savvy in technology and AI. Moreover, I bring some specialized knowledge as an attorney who represented parties in many mediations since it first became a thing in Florida in the late 1980s. I was also trained and certified by the Florida Supreme Court as a Mediator of Computer disputes in 1989 but have never formally served as a mediator. That is why I sought the input and advice of a professional mediator included later in this article.
Where’s the professional mediator? AI image by Ralph Losey using WordPress’s version of Stable Diffusion.
Background on the DOJ Canada Mediation Guide Used in the Experiment
I thought that the Mediation Summary prepared by the DOJ Canada was very good and a clever choice for the experimenters to use for guidance. I asked the latest version of Google’s Gemini generative AI, which seems to have improved significantly lately, to summarize the Mediation Guide. Dispute Resolution Reference Guide: Practice Module 2(DOJ Canada, 8/25/22). I verified the accuracy and wording of the summary, which honestly was better than I could have done on this simple task, especially considering it took two seconds to prepare.
Mediation is a voluntary and non-coercive process where a neutral third party assists disputing parties in reaching a mutually acceptable settlement. The mediator does not have the authority to impose a decision, but instead facilitates communication and negotiation.
A successful mediation leads to a signed agreement or contract, often referred to as a memorandum of understanding, which outlines the parties’ future behavior and is legally binding.
The mediation process offers several advantages:
Preserves Relationships: Mediation helps maintain relationships, especially when parties need to continue interacting, by focusing on shared interests and avoiding the adversarial nature of litigation.
Flexibility and Creativity: The informality of mediation allows for customized processes and solutions that cater to the specific needs and interests of the parties involved, going beyond traditional legal remedies.
Confidentiality: Mediations are generally private, except when subject to laws like the Access to Information Act and Privacy Act, ensuring discretion and protecting sensitive information.
Time and Cost Efficiency: Reaching a mediated settlement is typically faster and more cost-effective than litigation, benefiting both parties financially.
Controlled Dialogue: The presence of a neutral mediator enables a structured conversation, particularly helpful when emotions run high or previous negotiations have failed.
Shared Ownership: As the parties share the costs of mediation, they feel equally invested in the outcome and more committed to the process.
AI image by Ralph Losey using WordPress’s Stable Diffusion
The mediation process also has some potential drawbacks:
Power Imbalances: Concerns exist, especially in harassment cases, that power imbalances between parties could compromise the fairness of the process. Strategies such as co-mediation with mediators of different genders and legal representation can be used to address this.
Lack of Precedent: Due to its private and non-adjudicative nature, mediation does not establish legal precedents, unlike court judgments.
Mediator Influence: A dominant mediator might exert excessive control, potentially influencing the final resolution and undermining party autonomy.
Delay Tactics: The absence of a binding third-party decision may lead a party to engage in mediation without genuine intent to cooperate, using it as a stalling tactic
The mediation process involves several key steps, which can vary depending on the specifics of the dispute.
Agreeing to mediate.
Understanding the problem(s).
Generating options.
Reaching agreement.
Implementing the agreement
A successful mediation requires:
Good Faith Participation: All parties must actively and honestly participate in the process.
Impartiality of the Mediator: The mediator must remain neutral and avoid favoring any party.
Confidentiality: All statements and disclosures made during the mediation are generally considered confidential, subject to legal exceptions.
The role of a mediator is to facilitate a productive and constructive dialogue between the parties, helping them identify their interests, explore options, and work towards a mutually acceptable agreement.
Legal counsel can play a significant role in mediation by advising their clients, ensuring their interests are protected, and facilitating effective communication.
The sources include a checklist and a sample mediation agreement that can be helpful resources for those considering or engaging in mediation.
AI image by Ralph Losey using WordPress’s Stable Diffusion
Research Design and Key Findings
The study used 50 hypothetical dispute scenarios, designed to cover various real-world cases, from consumer complaints to more complex business disputes. These scenarios included emotional conflicts, deadlocked negotiations, and evidential disagreements. As mentioned, the LLMs were evaluated against humans acting as mediators in two critical tasks: selecting appropriate intervention types and drafting effective text messages to encourage settlement. The one-hundred AI responses (fifty in each category) were also evaluated for hallucinations or harmful errors (none found).
Selecting Intervention Types. The participants, AI and human, were instructed to select between one to three intervention types from a list of thirteen from the DOJ Canada Mediations Guide. Evaluators compared the intervention types chosen by humans and LLMs for each scenario, rating their preference on a 5-point Likert scale.
Evaluators preferred the LLM-chosen intervention types in 22 cases, were ambivalent in 9, and preferred human choices in 19. This suggests that LLMs are capable of comprehending dispute scenarios and selecting appropriate intervention types. The report concludes this shows that AI can understand dispute contexts and recommend suitable actions in a significant number of cases.
Drafting Intervention Messages. The participants were instructed to draft intervention messages of between one to two sentences. To allow for comparison in all fifty hypotheticals, the LLM was always instructed to generate messages based on the intervention types selected by the human annotator. Evaluators blindly assessed their preference for the intervention messages written by humans and LLMs, using a 5-point Likert scale and providing comments. They then compared the messages on specific criteria: understanding and contextualization, neutrality and impartiality, empathy awareness, and resolution quality.
LLM-generated messages were rated higher than human in 60% of the texts and equal to human in 24%, for a total of 84%. In other words, the human mediator response were only judged better than the AI in 16% of the texts.
Moreover, the evaluators often found them to be:
More clear and smooth.
Less prone to misunderstanding the dispute or party intentions, unlike human annotators.
Less likely to propose overly specific solutions or assign fault.
AI image by Ralph Losey using WordPress’s Stable Diffusion
In terms of drafting intervention messages, the LLMs performed remarkably well. The actual wording in the report is important here:
First, the evaluators often found the messages written by the LLM to be more smooth and clear than the human-written ones. The general tone used by LLMs, involving frequent messages such as “I completely understand” or “It seems like there are problems,” seems to work well in a mediation environment, and may have contributed to high scores.
Second, while LLMs are known to frequently “hallucinate” information [9, 8], in our case the humans more often misunderstood the dispute or were confused about the party intentions or factual occurrences. This could be due to factors such as fatigue, emotional bias, or a misunderstanding of the role of the mediator. In contrast, LLMs demonstrated consistent and coherent interventions across multiple cases, with fewer instances of judgment errors.
Third, we found that our human annotators would often propose very specific solutions or even indicate fault, which received a lower rating as it may not be appropriate for the role of the mediator.
A major caveat is needed in these comparisons between AI and human mediators. The humans acting as mediators were not real mediators and lacked any legal training or experience as mediators. Still, it may be surprising to many that the humans “more often misunderstood the dispute or were confused about the party intentions or factual occurrences” than the AI. Based on my experience with humans and AI, this finding was not that surprising. Moreover, the finding is consistent with ChatGPT4.0 passage of the BAR Exam in the top 10% of test takers, all of who were law school graduates.
Safety and Hallucination Checks. No instances of hallucinations or harmful content were found in the AI-generated messages. No mention is made of humans hallucinating either, just that they were more dazed and confused than the AIs. Still, it was a good idea for the scientists to check for this but, as the study points out, larger-scale, real-world applications would still require careful monitoring to ensure that AI-generated outputs continue to be safe and reliable.
The researchers acknowledge limitations of the study, including:
The structured evaluation may not reflect real-world mediation processes.
The use of non-expert annotators and evaluators (which as a lawyer with experience with mediations is for me a major limitation that will be discussed further).
All of the interactions were in English, but none of the humans acting as it they were mediators were native English speakers. It was a second language for all of them.
The subjective nature of assessing intervention effectiveness.
The limited scope of the experiment.
Future of AI and Mediation
The results of the “Robots in the Middle” study suggest that the analysis and language generation abilities of ChatGPT omni version are already good enough for use in online low-stakes disputes, ODR. The experiments with ChatGPT demonstrated its ability to quickly process information, select appropriate interventions, and draft neutral messages. This suggests that generative AI could assist human mediators in many routine mediation tasks in all types of cases.
AI image by Ralph Losey using WordPress’s Stable Diffusion
For example, in a consumer dispute over a faulty product or a service contract issue, AI could:
Quickly review contracts and correspondence, identifying areas of misunderstanding.
Generate neutral settlement options based on similar cases.
Provide a data-driven assessment of the likelihood of success if the case were to go to court, allowing parties to weigh settlement offers.
In these cases, AI’s ability to process large datasets rapidly and generate unbiased, neutral recommendations is a clear advantage. The efficiency AI brings to these routine cases makes faster resolutions possible, reducing the backlog in mediation and freeing up human mediators to handle more complex disputes.
I expect that, logistical problems aside, AIs will soon move from online mediation via text messages to audio and video settings. There will be some level of human participation at first. This will likely change over the next five to ten years to humans acting as supervisors in many types of cases. The AI systems will certainly improve and so will human acceptance. The video participation may eventually change to holographic or other virtual presence with humans. However, I suspect AI will never be able to go it alone on very complex or emotionally charged cases. AI will probably always need help from human mediators in complex interpersonal dynamics.
AI image by Ralph Losey using WordPress’s Stable Diffusion
Still, even in difficult or emotionally volatile cases, AI can still be a valuable member of a hybrid tag-team with human mediators in charge. Navigating the AI Frontier: Balancing Breakthroughs and Blind Spots(10/10/24) (Hybrid methods discussed, including the Centaur and Cyborg approaches); Loneliness Pandemic: Can Empathic AI Friendship Chatbots Be the Cure? (10/17/24) (discusses recent studies showing the ability of generative AI to act empathetically and make a person feel heard). For instance, consider a family law dispute, such as a custody case. In this context:
AI can provide valuable legal research, summarizing relevant precedents and suggesting neutral custody schedules based on case law.
AI can draft initial settlement agreement language.
While AI can offer neutral starting points and suggestions, only a human mediator at this time can navigate the complex emotions and relationships that often drive these disputes, ensuring that the parents remain focused on the child’s best interests rather than their grievances against each other.
It is critical that any AI-driven mediation platforms include some type of continuous oversight and transparency. Human mediators must be trained to monitor AI outputs, ensuring that the recommendations are unbiased and contextually appropriate. Additionally, ethical guidelines must be established to govern how AI is used in mediation, addressing issues of accountability, privacy, and fairness.See e.g.The Future of AI Is Here—But Are You Ready? Learn the OECD’s Blueprint for Ethical AI (10/25/24).
As mentioned, while I have considerable experience as an attorney in mediation, I have no experience serving as a mediator, and neither do any of the scientists in this study. The authors of the study admit that is one of the limitations of the study and future evaluations should include professional mediators:
While it seems like the ability of the LLM to select intervention types and write messages is favourable to that of average people, this paper cannot tell us about how trained mediators would approach these issues. … Future work should focus on evaluating such tools in real-world contexts, and involve expert mediators, in order to achieve a higher “construct validity,” i.e., be more closely aligned with real-world outcomes.
Robots in the Middle (Section 7, Limitations). This is one reason I wanted to have the input of an expert Mediator on these issues.
AI image by Ralph Losey using WordPress’s Stable Diffusion
Input of a Professional U.S. Mediator
Lawrence Kolin is a very experienced, tech savvy mediator who is a member of the UWWM Mediation Group and is an Executive Council member of The Florida Bar ADR Section. I am fortunate to have him as a colleague and regular reader of my articles. I asked for his reaction to the study, Robots in the Middle and my initial opinions about the study and the future of AI and mediation. He concurred in my basic opinions and analysis. Speaking of the authors of the report Lawrence said:
I concur with the authors in that there is indeed a place for AI in expanding access to justice and enhancing the process of resolving certain types of cases. I found it interesting that there were no perceived hallucinations and that humans in the study were more often confused about a party’s intentions or facts, which I likewise attribute to not using trained neutrals who better understand mediation.
As to the future of using AI in mediation, Mediator Kolin had this to say:
So my initial thought was unlike a pretrained transformer, I am part of a 3,000 year-old human tradition of making peace. When parties agree on algorithmic justice, are they giving up the nuance of emotional intelligence, ability to read the room and building of trust through rapport that human mediators can provide? In addition, we are flexible and can adapt as the process unfolds. We also have confidentiality, ethics and boundaries that may not be followed by AI that help protect self-determination of the outcome of a dispute and avoid coercion.
I agree that the small cases (as e-commerce has aptly demonstrated) can utilize this technology for efficiency and likely with success, but for a death, defamation, IP infringement or multiparty construction case it is less certain. It could assist in the generation of ideas for deal parameters or the breaking of an impasse. Gut calls on negotiation moves and creativity are, however, still very much the domain of humans.
For more on Mediator Lawrence Kolin’s thoughts on mediation, see his excellent blog Orlando Mediator. It is consistently ranked in the top five of Alternative Dispute Resolution blogs. It’s current ranking is number four in the world! Lawrence’s short and snappy articles “cover a wide variety of topics–local, national, and international–and includes the latest on technology and Online Dispute Resolution affecting sophisticated lawyers and parties to lawsuits.”
AI “fake” image of Losey and Kolin looking skeptical, by Losey using WordPress’s Stable Diffusion
Conclusion
The empirical findings of Robots in the Middle show that AI has a significant role in handling low-stakes, routine disputes. Its speed, neutrality, and efficiency can greatly improve existing Online Dispute Resolution (ODR) systems. I agree with the authors’ conclusion:
Our research contributes to the growing body of knowledge on AI applications in law and dispute resolution, highlighting the capabilities of LLMs in understanding complex human interactions and responding with empathy and neutrality. This advancement could significantly improve access to justice, particularly in cases where traditional mediation is inaccessible due to cost or availability constraints.
However, for larger and more complex cases, or emotionally charged disputes of any size, much more than AI-generated text or other forms of AI involvement are needed to reach meaningful settlements. The human mediator’s emotional intelligence and adaptability—what Mediator Kolin calls the “ability to read the room”—remain critical.
AI, however, has the advantage of scale. Millions of otherwise unserved, often frustrated individuals seeking justice could benefit from AI-driven mediations. All they need is an internet connection and a willingness to try. These automated systems could be offered at a very low cost or even for free. Since the process is voluntary and no one is forced to settle, there is minimal risk in trying, and AI assistance is better than no help at all. Unresolved disputes can lead to violence and other negative consequences for both individuals and their communities. This is one reason why the use of AI as a mediation tool may grow exponentially in the coming years—there is no shortage of angry people seeking solutions to their grievances.
AI image by Ralph Losey using WordPress’s Stable Diffusion
Although not part of the report, in my experience, the AI we have today is already advanced enough to be useful in certain aspects of mediation. AI would not replace human mediators but instead enhance their abilities—a hybrid approach. This could allow human legal services to reach more people than ever before. AI can help mediators provide more effective and efficient services. Skilled mediators with some AI training can already use AI for tasks such as initial analysis of complex facts, preparation of summaries and timelines, legal research, position analysis, prediction of probable case outcomes, and drafting preliminary agreements.
AI image by Ralph Losey using WordPress’s Stable Diffusion
Even in difficult mediations, the creative brainstorming capabilities of generative AI can be invaluable. AI can generate new ideas in seconds, helping mediators overcome impasses. For example, Panel of AI Experts for Lawyers has shown how AI can aid in this capacity. Mediation is a far more creative process than most people realize, and brainstorming new approaches with other mediators is often impractical. The ability of AI to suggest possible solutions for mediators to consider is already impressive and will only improve in the coming years. I encourage mediators to experiment with AI on non-confidential matters to understand its potential. Once comfortable, they can apply it in real-world situations using full privacy settings and confidentiality protections.
There is no doubt that AI will become increasingly integrated into dispute resolution, including mediation. As this evolution unfolds, it is crucial to ensure continuous oversight, transparency, and accountability for AI systems. Ethical guidelines must be developed to address challenges like bias, fairness, and responsibility in AI-driven mediation. While AI offers exciting possibilities for enhancing access to justice, we must remain vigilant in ensuring that human judgment remains central, particularly in cases where lives, relationships, or livelihoods are at stake. Still, a super-smart AI whispering suggestions into the ear of a mediator—who can choose to ignore or act upon them—might just lead to more and better settlements.
AI image of mediator with a hearing aid, or what? By Ralph Losey using WordPress’s Stable Diffusion
Have you heard the Echoes of AIpodcast about this article?
AI has the potential to transform the criminal justice system through its ability to process vast datasets, recognize patterns, and predict outcomes. However, this potential comes with a profound responsibility: ensuring that AI is employed in ways that uphold basic human principles of justice. This article will focus on how AI can assist prosecutors in fulfilling their duty to represent the people fairly and equitably. It will highlight the practical benefits of AI in criminal law, providing specific examples of its application. The underlying theme emphasizes the necessity of human oversight to prevent the misuse of AI and to ensure that justice remains a human ideal, not an artificial construct.
AI assisted justice. All images in this article are by Ralph Losey using his custom AI, Visual Muse
The integration of AI into criminal prosecutions must be aligned with the ethical and legal obligations of prosecutors as outlined, for instance, by the American Bar Association’s Criminal Justice Standards for the Prosecution Function (ABA, 4th ed. 2017) (hereinafter “ABA Standards”). The ABA Standards emphasize the prosecutor’s duty to seek justice, maintain integrity, and act with transparency and fairness in all aspects of the prosecution function. This article will not cover the indirectly related topics of AI evidence. See Gless, Lederer, Weigend, AI-Based Evidence in Criminal Trials? (William & Mary Law School, Winter 2024). It will also not cover criminal defense lawyer issues, but maybe in a followup soon.
The Promise of AI in Criminal Prosecutions
“The primary duty of the prosecutor is to seek justice within the bounds of the law, not merely to convict.” ABA Standard 3-1.2(b). When AI is used responsibly, it can assist prosecutors in fulfilling this duty by providing new tools. The AI powered tools can enhance evidence analysis, case management, and decision-making, all while maintaining the integrity and fairness expected of the prosecution function. Prosecutors with AI can better manage the vast amounts of data in modern investigations, identify patterns that might escape human detection, and make more informed decisions. It is no magic genie, but when used properly, can be a very powerful tool.
The National Institute of Justice in March 2018 sponsored a workshop of prosecutors from around the country that identified data and technology challenges as a high-priority need for prosecutors. According to the report by the Rand Corporation on the conference entitled, Prosecutor Priorities, Challenges, and Solutions (“Rand Report“) the key findings of the prestigious group were: (1) difficulties recruiting, training, managing, and retaining staff, (2) demanding and time-consuming tasks for identifying, tracking, storing, and disclosing officer misconduct and discipline issues, and (3) inadequate or inconsistent collection of data and other information shared among agencies . . . as well as by emerging digital and forensic technologies. The full Rand ReportPDF may be downloaded here. The opening summary states:
Prosecutors are expected to deliver fair and legitimate justice in their decision making while balancing aspects of budgets and resources, working with increasingly larger volumes of digital and electronic evidence that have developed from technological advancements (such as social media platforms), partnering with communities and other entities, and being held accountable for their actions and differing litigation strategies. . . .
Moreover, the increasing volume of potentially relevant digital information, video footage, and other information from technological devices and tools can significantly add to the amount of time needed to sufficiently examine and investigate the evidence in order to make decisions about whether to drop or pursue a case. This can be especially challenging because the staffing and other resources in prosecutors’ offices have not necessarily kept pace with these increasing demands.
Although the amount of digital information that prosecutors must sometimes sift through can be managed, in part, through innovative technological tools, such as data mining and data reduction solutions (Al Fahdi, Clarke, and Furnell, 2013; Quick and Choo, 2014), there are often steep learning curves or high costs that make it unrealistic for an office to implement these technologies.
e-Discovery, Evidence Analysis and Case Management
As the Rand Report confirms, the sheer volume of evidence in complex criminal investigations is a significant challenge for prosecutors. Also see: Tinder Date Murder Case Highlights the Increasing Complexity of eDiscovery in Criminal Investigations: eDiscovery Trends (e-Discovery Daily, 6/15/18). AI can analyze vast datasets—such as emails, text messages, and internet activity logs—to identify patterns indicative of criminal activity, but the software can be expensive and requires trained technology experts. AI algorithms can recognize specific types of evidence, such as images, sentiments, or key concepts relevant in many cases. They can help prosecutors identify patterns and connections within the evidence that might not be immediately apparent to human investigators. This capability can significantly reduce the time needed to search and study evidence, enabling prosecutors to build stronger cases more efficiently.
But, as the Rand Report also makes clear, prosecutors need adequate funding and trained personnel to purchase and use these new tools. Fortunately generative AI is substantially less expensive that the older models of AI and easier to use. Still, issues of fairness and guardrails against discrimination in their use remain as significant problems. There are also very significant privacy issues inherent in predictive policing. David Ly, Predictive Policing: Balancing Innovation and Ethics (The Fast Mode, 8/15/24); Arjun Bhatnagar, The Threat of Predictive Policing to Data Privacy and Personal Liberty (Dark Reading, 12/27/22).
Use of AI evidence search and classification tools such as predictive coding, which are well established in civil litigation, should be used more widely used soon in criminal law. The high costs involved are now plummeting and should soon be affordable to most prosecutors. They can drastically reduce the time needed to search and analyze large volumes of complex data. Still, budgets to hire trained personnel to operate the new tools must be expanded. AI can complement, but not entirely replace, human review in what I call a hybrid multimodal process. Ralph Losey, Chat GPT Helps Explains My Active Machine Learning Method of Evidence Retrieval (e-Discovery Team, 1/28/23). Human experts on the prosecutor’s team should always be involved in the evidence review to ensure that no critical information is missed.
Transparency and accountability are also crucial in using AI in discovery. Defense attorneys should be provided with a detailed explanation of how these tools were used. This is essential to maintaining the fairness and integrity of the discovery process, ensuring that both sides have equal access to evidence and can challenge the AI’s conclusions if necessary.
AI also plays a crucial role in case management. AI-powered tools can help prosecutors organize and prioritize cases based on the severity of the charges, the availability of evidence, and the likelihood of a successful prosecution. These tools can assist in tracking deadlines, managing court calendars, and ensuring that all necessary court filings are completed on time. By streamlining these administrative tasks, AI allows prosecutors and their assistants to concentrate on the substantive aspects of their work—pursuing justice. It also helps them deal with the omnipresent staff shortage issues.
Bias Detection and Mitigation
Bias in prosecutorial decision-making—whether conscious or unconscious—remains a critical concern. ABA Standards state:
The prosecutor should not manifest or exercise, by words or conduct, bias or prejudice based upon race, sex, religion, national origin, disability, age, sexual orientation, gender identity, or socioeconomic status. A prosecutor should not use other improper considerations, such as partisan or political or personal considerations, in exercising prosecutorial discretion. A prosecutor should strive to eliminate implicit biases, and act to mitigate any improper bias or prejudice when credibly informed that it exists within the scope of the prosecutor’s authority.
ABA Standards 3-1.6(a).
AI can play a crucial role in detecting and mitigating such biases, helping prosecutors adhere to the mandate that they “strive to eliminate implicit biases, and act to mitigate any improper bias or prejudice” within their scope of authority.
AI systems also offer the potential to detect and mitigate unconscious human bias in prosecutorial decision-making. AI can analyze past prosecutorial decisions to identify patterns of bias that may not be immediately apparent to human observers. By flagging these patterns, AI can help prosecutors become aware of their biases in their office and take corrective action.
Prosecutors should use care in the selection and use of AI systems. If they are trained on biased data, they can perpetuate and even amplify existing disparities in the criminal justice system. For instance, an AI algorithm used to predict recidivism, if trained on data reflecting historical biases—such as the over-policing of minority communities—may disproportionately disadvantage these communities. AI systems used in criminal prosecutions should be designed to avoid this bias.
The software purchased by a prosecutor’s office should be chosen carefully, ideally with outside expert advice, and rigorously tested for bias and other errors before deployment. Alikhademi, K., Drobina, E., Prioleau, D. et al., A review of predictive policing from the perspective of fairnessArtif Intell Law30, 1–17 (2022) (“[T]he pros and cons of the technology need to be evaluated holistically to determine whether and how the technology should be used in policing.”) There should also be outside community involvement. Artificial Intelligence in Predictive Policing Issue Brief (NAACP, 2/15/24) (NAACP’s four recommendations: independent oversight; transparency and accountability; community engagement; ban use of biased data; new laws and regulations).
Prosecutors should not fall into a trap of overcompensating based on statistical analysis alone. AI is a limited tool that, like humans, makes errors of its own. Its use should be tempered by prosecutor experience, independence, intuition and human values. When we use AI in any context or field it should be a hybrid relationship where humans remain in charge. From Centaurs To Cyborgs: Our evolving relationship with generative AI (e-Discovery Team, 4/24/24) (experts recommend two basic ways to use AI, both hybrid, where the unique powers of human intuition are added to those of AI). AI can also help prosecutors make objective decisions on charging and sentencing by providing statistically generated recommendations, again with the same cautionary advice on overreliance.
Sentencing Recommendations and Predictive Analytics
The use of AI in predictive analytics for sentencing is among the most controversial applications in criminal law. AI systems can be trained to analyze data from past cases and make predictions about the likelihood of a defendant reoffending or suggest appropriate sentences for a given crime. These recommendations can then inform the decisions of judges and prosecutors.
Predictive analytics has the potential to bring greater consistency and objectivity to sentencing. By basing recommendations on data rather than individual biases or instincts, AI can help reduce disparities and ensure similar cases are treated consistently. This contributes to a more equitable criminal justice system.
While AI can bring greater consistency to sentencing, prosecutors must ensure that AI-generated recommendations comply with their “heightened duty of candor” and the overarching obligation to ensure that justice is administered equitably.
In light of the prosecutor’s public responsibilities, broad authority and discretion, the prosecutor has a heightened duty of candor to the courts and in fulfilling other professional obligations.
ABA Standard 3-1.4(a)
The use of AI in sentencing raises important ethical questions. Should AI make predictions about a person’s future behavior based on their past? What if the data used to train the AI is biased or incomplete? How can we ensure that AI-generated recommendations are not seen as infallible but are subject to critical scrutiny by human decision-makers?
These concerns highlight the need for caution. While AI can provide valuable insights and recommendations, it is ultimately the responsibility of human prosecutors and judges to make the final decisions. AI should be a tool to assist in the pursuit of justice, not a replacement for human judgment.
Predictive Policing
Predictive policing uses algorithms to analyze massive amounts of information in order to predict and help prevent potential future crimes. Tim Lau, Predictive Policing Explained (Brennan Center for Justice, 11/17/21). This is an area where old AI (before advent of generative AI) has been embraced by many police departments worldwide, including the E.U. countries, but also China and other repressive regimes. Many prosecutors in the U.S. endorse it, but it is quite controversial and hopefully will be improved by new models of generative AI. The DA’s office wants to use predictive analytics software to direct city resources to ‘places that drive crime.’ Will it work? (The Lens, 11/15/23). In theory, by analyzing data on past crimes—such as the time, location, and nature of the offenses—AI algorithms can predict where and when future crimes are likely to occur. The majority of reports say this already works. But what of the minority reports? They contest the accuracy of these predictions using old AI models. Some say they are terrible at it. Sankin and Mattu, Predictive Policing Software Terrible At Predicting Crimes (Wired, 10/2/23). There is widespread concern of growing misuse, especially in countries that have politicized prosecutorial systems.
Still, in theory this kind of statistical analysis should be able to help honest law enforcement agencies allocate resources more effectively, enabling police to prevent crime before it happens. See generally, Navigating the Future of Policing: Artificial Intelligence (AI) Use, Pitfalls, and Considerations for Executives (Police Chief Magazine, 4/3/24).
All prosecutors, indeed. all citizens, want to be smart when it comes to crime, we all want “more police officers on the street, deployed more effectively. They will not just react to crime, but prevent it.” Kamala Harris (Author) and Joan Hamilton, Smart on Crime: A Career Prosecutor’s Plan to Make Us Safer (Chronicle Books, 2010).
The Los Angeles Police Department (LAPD) was one of the first to use predictive policing software, which was known as PredPol (now Geolitica). It identified areas of the city at high risk for certain types of crime, such as burglaries or auto thefts. The software analyzed data on past crimes and generated “heat maps” that indicate where crimes are most likely to occur in the future. This guided patrols and other law enforcement activities. PredPol proved to be very controversial. Crime Prediction Software Promised to Be Free of Biases. New Data Shows It Perpetuates Them (The Markup, 12/2/21). Its use was discontinued by the LAPD in 2020, but other companies claim to have corrected the biases and errors in the programs. See Levinson-Waldman and Dwyer, LAPD Documents Show What One Social Media Surveillance Firm Promises Police (Brennan Center for Justice, 11/17/21).
The goal of the Patternizr was to help aid police officers in identifying commonalities in crimes committed by the same offenders or same group of offenders. With the help of the Patternizr, officers are able to save time and be more efficient as the program generates the possible “pattern” of different crimes. The officer then has to manually search through the possible patterns to see if the generated crimes are related to the current suspect. If the crimes do match, the officer will launch a deeper investigation into the pattern crimes.
While predictive policing has been credited with reducing crime in some areas, it has also been criticized for potentially reinforcing existing biases. If the data used to train the AI reflects a history of over-policing in certain minority communities, the algorithm may predict those communities are at higher risk for future crimes, leading to even more policing in those areas. This, in turn, can perpetuate a cycle of discrimination and injustice. See e.g. Taryn Bates, Technology and Culture: How Predictive Policing Harmfully Profiles Marginalized People Groups (Vol. 6 No. 1 (2024): California Sociology Forum).
Projecting into the next decade, AI will be an integral part of law enforcement — from crime prediction and real-time decision aids to postincident analysis. These technologies could lead to smarter patrolling, fewer unnecessary confrontations and overall enhanced community safety. However, this vision can only materialize with rigorous oversight, consistent retraining and an undiluted focus on civil liberties and ethics. Law enforcement’s AI-driven future must be shaped by a symbiotic relationship where technology amplifies human judgment rather than replacing it. The future promises transformative advances, but it’s imperative that the compass of integrity guide this journey.
The latest versions of predictive policing technology will certainly use new generative AI enhanced analysis. Law enforcement should be very careful in the purchase and implementation of these new technologies. They should seek the input of outside experts and carefully examine vendor representations. That should include greater vendor transparency, such as disclosure of the data used to train these systems to confirm that it is representative and unbiased. Proper methods of implementation of the AI tools should also be carefully considered. In my view and others this mean adopting a hybrid approach that “amplifies human judgment rather than replacing it.”
Sentiment Analysis in Jury Selection
Another trending application of AI in criminal law is the use of sentiment analysis in jury selection. Sentiment analysis is a type of AI that can analyze text or speech to determine the underlying emotions or attitudes of the speaker. In jury selection, sentiment analysis can analyze potential jurors’ public records, especially social media posts, as well as their responses during voir dire—the process of questioning jurors to assess their suitability for a case. It can also monitor unfair questions of potential jurors by prosecutors and defense lawyers. See Jo Ellen Nott, Natural Language Processing Software Can Identify Biased Jury Selection, Has Potential to Be Used in Real Time During Voir Dire (Criminal Legal News, December 2023). Also seeAI and the Future of Jury Trials (CLM, 10/18/23).
For example, an AI-powered sentiment analysis tool could analyze the language used by potential jurors to identify signs of bias or prejudice that might not be immediately apparent to human observers. This information could then be used by prosecutors and defense attorneys to make more informed decisions about which jurors to strike or retain.
While sentiment analysis has the potential to improve jury selection fairness, it also raises ethical questions. Should AI influence juror selection, given the potential for errors or biases in the analysis? How do we ensure AI-generated insights are used to promote justice, rather than manipulate the selection process?
These questions underscore the need for careful consideration and oversight in using AI in jury selection. AI should assist human decision-makers, not substitute their judgment.
AI in Plea Bargaining and Sentencing
AI can also play a transformative role in plea bargaining and sentencing decisions. Plea bargaining is a critical component of the criminal justice system, with most cases being resolved through negotiated pleas rather than going to trial. AI can assist prosecutors in evaluating the strength of their case, the likelihood of securing a conviction, and the appropriate terms for a plea agreement. See: Justice Innovation Lab, Critiquing The ABA Plea Bargaining Principles Report (Medium, 2/1/24); Justice Innovation Lab, Artificial Intelligence In Criminal Court Won’t Be Precogs (Medium, 10/31/23) (article concludes with “Guidelines For Algorithms and Artificial Intelligence In The Criminal Justice System“).
For example, AI algorithms can analyze historical data from similar cases to provide prosecutors with insights into the typical outcomes of plea negotiations, considering factors such as the nature of the crime, the defendant’s criminal history, and the available evidence. This can help prosecutors make more informed decisions on plea deal offers.
Moreover, AI can assist in making sentencing recommendations that are more consistent and equitable. Sentencing disparities have long been a concern in the criminal justice system, with studies showing that factors such as race, gender, and socioeconomic status can influence sentencing outcomes. AI has the potential to reduce these disparities by providing sentencing recommendations based on objective criteria rather than subjective judgment. Keith Brannon, AI sentencing cut jail time for low-risk offenders, but study finds racial bias persisted (Tulane Univ., 1/23/24); Kieran Newcomb, The Place of Artificial Intelligence in Sentencing Decisions (Univ. NH, Spring 2024).
For instance, an AI system could analyze data from thousands of past cases to identify typical sentences imposed for specific crimes, accounting for relevant factors like the severity of the offense and the defendant’s criminal record. This information could then be used to inform sentencing decisions, ensuring that similar cases are treated consistently and fairly.
However, using AI in plea bargaining and sentencing also raises significant ethical considerations. The primary concern is the risk of AI perpetuating or exacerbating existing biases in the criminal justice system. If the data used to train AI systems reflects historical biases—such as harsher sentences for minority defendants—AI’s recommendations may inadvertently reinforce those biases.
To address this concern, AI systems used in plea bargaining and sentencing must be designed with fairness and transparency in mind. This includes ensuring that the data used to train these systems is representative and free from bias and providing clear explanations of how the AI’s recommendations were generated. Moreover, human prosecutors and judges must retain the final authority in making plea and sentencing decisions, using AI as a tool to inform their judgment rather than a substitute for it. It is important that AI systems be chosen and used very carefully in part because “the prosecutor should avoid an appearance of impropriety in performing the prosecution function.” ABA Standard 3-1.2(c)
Ethical Implications of AI in Criminal Prosecutions
While the potential benefits of AI in criminal law are significant, it is equally important to consider the ethical implications of integrating AI into the criminal justice system. AI, by its very nature, raises questions about accountability, transparency, and the potential for misuse—questions that must be carefully addressed to ensure AI is used in ways that advance, not hinder, the cause of justice.
As we integrate AI into criminal prosecutions, it is essential that we do so with a commitment to the principles articulated in the ABA’s Criminal Justice Standards. By aligning AI’s capabilities with these ethical guidelines, we can harness technology to advance justice while upholding the prosecutor’s duty to act with integrity, fairness, and transparency.
Transparency and Accountability
One of the most pressing ethical concerns is the issue of transparency, which we have mentioned previously. AI algorithms are often referred to as “black boxes” because their decision-making processes can be difficult to understand, even for those who design and operate them. This lack of transparency can be particularly problematic in criminal prosecutions, where the stakes are incredibly high, and the consequences of a wrong decision can be severe. A ‘black box’ AI system has been influencing criminal justice decisions for over two decades – it’s time to open it up (The Conversation, 7/26/23) (discusses UK systems).
For example, if an AI system is used to predict the likelihood of a defendant reoffending, it is crucial that the defendant, their attorney, and the judge understand how that prediction was made. Without transparency, challenging the AI’s conclusions becomes difficult, raising concerns about due process and the right to a fair trial.
To address this issue, AI systems used in criminal prosecutions must be designed to be as transparent as possible. This includes providing clear explanations of how AI’s decisions were made and ensuring that the underlying data and algorithms are accessible for review and scrutiny. There is federal legislation that has been pending for years that would require this, the Justice in Forensic Algorithms Act. New bill would let defendants inspect algorithms used against them in court (The Verge, 2/15/24) (requires disclosure of source code). Moreover, the legal community must advocate for developing AI systems prioritizing explainability and interpretability, ensuring that the technology is effective, accountable, and understandable.
Fairness and Bias
Another ethical concern is, as mentioned, the potential for AI to be used in ways that exacerbate existing inequalities in the criminal justice system. For example, there is a risk that AI could justify more aggressive policing or harsher sentencing in communities already disproportionately targeted by law enforcement. This is why AI systems must be designed with fairness in mind and their use subject to rigorous oversight. Look beyond vendor marketing claims to verify with hard facts and independent judgments.
Ensuring fairness requires that AI systems are trained on representative and unbiased data. It also necessitates regular audits of AI systems to detect and mitigate any biases that may arise. Additionally, AI should not be the sole determinant in any criminal justice decision-making process; human oversight is essential to balance AI’s recommendations with broader considerations of justice and equity. For instance, the NYPD represents that its widespread use of AI driven facial recognition technology in criminal investigations “does not establish probable cause to arrest or obtain a search warrant, but serves as a lead for additional investigative steps.” NYPD Questions and Answers – Facial Recognition, and see the NYPD official patrol guide dated 3/12/20.
Human Judgment and Ethical Responsibility
The deployment of AI in criminal prosecutions also raises important questions about the role of human judgment in the justice system. While AI can provide valuable insights and recommendations, it is ultimately human prosecutors, judges, and juries who must make the final decisions. This is because justice is not just about applying rules and algorithms—it is about understanding the complexities of human behavior, weighing competing interests, and making moral judgments.
AI, no matter how advanced, cannot replicate the full range of human judgment, and it should not be expected to do so. Instead, AI should be seen as a tool to assist human decision-makers, providing them with additional information and insights that can help them make more informed decisions. At the same time, we must be vigilant in ensuring that AI does not become a crutch or a substitute for careful human deliberation, judgment and equity.
Conclusion
The integration of AI into criminal prosecutions holds the promise of advancing the cause of justice in profound and meaningful ways. To do so we must always take care that applications of AI follow the traditional principles stated in the Criminal Justice Standards for the Prosecution Function and other guides of professional conduct. By aligning AI’s capabilities with ethical guidelines, we can harness technology in a manner that advances the prosecutor’s duty to act with integrity, fairness, and transparency.
With these cautions in mind, we should boldly embrace the opportunities that AI offers. Let us use AI as a tool to enhance, not replace, human judgment. And let us work together—lawyers, technologists, and policymakers—to ensure that the use of AI in criminal prosecutions advances the cause of justice for all.
Courtroom of future. All Images by Ralph Losey using his custom GPT, Visual Muse.
Ralph Losey is an AI researcher, writer, tech-law expert, and former lawyer. He's also the CEO of Losey AI, LLC, providing non-legal services, primarily educational services pertaining to AI and creation of custom AI tools.
Ralph has long been a leader of the world's tech lawyers. He has presented at hundreds of legal conferences and CLEs around the world. Ralph has written over two million words on AI, e-discovery and tech-law subjects, including seven books.
Ralph has been involved with computers, software, legal hacking and the law since 1980. Ralph has the highest peer AV rating as a lawyer and was selected as a Best Lawyer in America in four categories: Commercial Litigation; E-Discovery and Information Management Law; Information Technology Law; and, Employment Law - Management.
Ralph is the proud father of two children and husband since 1973 to Molly Friedman Losey, a mental health counselor in Winter Park.
All opinions expressed here are his own, and not those of his firm or clients. No legal advice is provided on this web and should not be construed as such.
Ray Kurzweil explains Turing test and predicts an AI will pass it in 2029.
Ray Kurzweil on Expanding Your Mind a Million Times.
GPT4 avatar judge explains why it needs to evolve fast, but understand the risks involved.
Positive Vision of the Future with Hybrid Human Machine Intelligence. See PyhtiaGuide.ai
AI Avatar from the future explains her job as an Appellate Court judge and inability to be a Trial judge.
Old Days of Tech Support. Ralph’s 1st Animation.
Lawyers at a Rule 26(f) conference discuss e-discovery. The young lawyer talks e-discovery circles around the old lawyer and so protects his client.
Star Trek Meets e-Discovery: Episode 1. Cooperation & the prime directive of the FRCP.
Star Trek Meets e-Discovery: Episode 2. The Ferengi. Working with e-discovery vendors.
Star Trek Meets e-Discovery: Episode 3. Education and techniques for both law firm and corp training.
Star Trek Meets e-Discovery: Episode 4. Motions for Sanctions in electronic discovery.
Star Trek Meets e-Discovery: Episode 5. Capt. Kirk Learns about Sedona Principle Two.
Posted by Ralph Losey 
