The Algorithmic Family

How AI Is Rewriting the Rules of Family Law

September/October 2024 By Maha Kamal

This article explores the opportunities and challenges of artificial intelligence in the practice of family law and discusses implications of this technology for clients, practitioners, and judges.

As we stand at the cusp of a technological revolution, the intersection of artificial intelligence (AI) and family law presents a paradigm shift, offering unprecedented opportunities and challenges for legal practitioners in Colorado and beyond. The emergence of AI-assisted case management and case law research, as well as the rise of ChatGPT in the public domain, is just the beginning. This article addresses the history of this technology, its potential benefits, and its consequences for major family law players like clients, practitioners, and judicial officers. It also considers potential future uses of AI in family law, explores how practitioners and judges can maintain the human element that makes family law unique among practice areas, and raises some of the critical ethical considerations that AI poses in this practice area.

AI: A (Very) Brief History

Artificial intelligence is not new. Since the 1970s, researchers such as Geoffrey Hinton—dubbed the “godfather” of AI—have experimented with ways computers could mimic the neural processes that give way to human intelligence. Hinton tinkered for decades, building bigger neural nets structured in ingenious ways.1 Over those years, computers got faster, and neural nets, drawing on data available on the internet, started transcribing speech, playing games, translating languages, and even driving cars.2 Now, computers are evolving at lightning speed, with neural networks projected to surpass human cognition within 10 years in a phenomenon known as the “singularity.”3

The AI we’re most familiar with today is a generative pre-trained transformer (GPT).4 This technology was developed by inputting large datasets into computers and then instructing them to break down this data into “tokens.” The trained computers then generate responses to user questions or requests by predicting which token will occur next based on tokens encountered in the datasets. Datasets include various sources, such as books and popular websites like Reddit and Wikipedia.

To the average user, AI chatbots like ChatGPT can be deceiving because they will generate fluent and complex responses, but that doesn’t mean the responses are always accurate. The GPT technology isn’t thinking, reasoning, planning, and problem solving the same way a human would. Instead, it’s fine-tuning a complex probability mechanism based on human data. This means that sometimes the data can be startlingly wrong.

AI’s appearance in the law is relatively new and has mostly made headlines where mistakes have been made. Perhaps the most famous case to date involving AI is Mata v. Avianca, Inc.5 In that case, New York plaintiff’s attorney Peter LoDuca was issued a show cause order after relying on ChatGPT to assist with a federal briefing in which “six of the submitted cases appear[ed] to be bogus judicial decisions with bogus quotes and bogus internal citations.”6 It cost Attorney LoDuca his public reputation and $5,000 in penalties and fees.7

On November 22, 2023, Colorado suspended a new attorney after he included nonexistent law in a motion in a civil matter that he drafted using ChatGPT. After he discovered the errors, he failed to alert the court of the deficiencies and instead falsely attributed the mistakes to his legal intern.8

Many aspects of family law practice remain dated, with limited docketing systems, antiquated courtrooms, and slow timelines. However, our practice is now encountering AI-generated parenting plans, pleadings, and substantial changes in parental communication with the rise of ChatGPT. As GPT technology and AI advance exponentially, we must better prepare the field for the future by considering how these developments affect clients, practitioners, and judicial developments.

Client Considerations

Family law litigants, whether spouses, parents, children, or nonparents, have distinct legal needs ranging from assistance with divorce and separation to allocating parental responsibilities. Their input is required for producing disclosures and preparing essential documents, including separation agreements and parenting plans. Their journey involves communicating with ex-partners and co-parents and interacting with attorneys and judges. These litigants represent a broad spectrum of socioeconomic backgrounds that influence their access to legal resources and shape their experiences in the legal system. Despite these differences, all share a common objective: formalizing familial separation and securing orders to move themselves and their families forward. AI can assist clients by making self-representation less daunting, providing parenting and financial tools, and offering coaching to help them cope with their emotions throughout their case.

Self-Representation

Upwards of 80% of Coloradans enter the family law system unrepresented.9 Many efforts, from pro bono initiatives like Metro Volunteer Lawyers to the new limited licensure program for legal paraprofessionals that launched in 2023, attempt to alleviate this substantial demand. AI provides another way for litigants to help themselves through an otherwise complex legal system. This innovative technology does not come without controversy—it presents a unique ethical question for state regulatory agencies. For example, at the request of the Access to Justice Commission, a Colorado Supreme Court advisory committee recently established a subcommittee to evaluate existing rules barring the unauthorized practice of law to determine if changes are necessary to accommodate emerging AI-powered legal tools.10 The commission expressed concern that the current unauthorized practice of law rules may prohibit new technologies in Colorado from being adopted for use in the legal system.11

However, AI technology is already readily accessible to the public without restriction. By educating self-represented litigants on how to use AI to understand their cases and navigate the legal system, we reduce the risk of misinformation and litigation abuse.

Technology such as ChatGPT offers seemingly endless benefits to self-represented litigants. This technology could analyze and explain essential court documents, such as case management or pretrial orders. This would better inform self-represented litigants about the expectations of the court and help them comply with orders concerning financial disclosures, parenting time orders, and discovery requests. Rather than appearing for court hearings or mediation unprepared and uncertain, AI could help prepare the self-represented litigant and establish reasonable expectations. Current technology could also assist with explaining and calculating basic child support or the allocation of assets and debts with equitable considerations. Whenever needed, AI could also encourage the self-represented litigant to seek legal counsel if a case appears too complicated.

AI-Powered Parenting and Financial Tools

There is vast potential for AI-driven parenting and financial tools. For example, AI could be integrated into current online tools commonly used by parent litigants, such as Civil Communicator and Family Law Software. Civil Communicator monitors and filters parent communications it deems inappropriate or unhelpful. AI could improve and streamline this service and add helpful components like coaching to help users learn to communicate effectively and respectfully. It could take a rejected message and explain to the parent how to rephrase their ask or response, a feature not currently available on the platform. The popular OurFamilyWizard platform recently incorporated an AI feature called ToneMeter™ that analyzes text and helps a parent compose messages that reduce conflict and miscommunication.12

Similarly, Family Law Software offers a limited free service through the Colorado Judicial Branch’s website that generates child support worksheets.13 Litigants often misunderstand or misread these documents or are unsure how to input data. AI could streamline this process by guiding the parents through a more interactive experience designed to reduce errors. For example, it could request that a parent’s gross income include employment and other qualified CRS § 14-10-115 sources and explain that work over 40 hours a week should not be included.

Practitioners today often submit asset/debt worksheets to courts in dissolution or other legal proceedings. These worksheets are available online to clients and self-represented litigants. AI could be incorporated into the Judicial Branch’s website to assist litigants with preparing these forms for the court in a meaningful and accurate way, whether they transfer data from a Sworn Financial Statement in one click or ask how a user would like to allocate a certain property or debt. AI could also analyze financial pleadings, such as the asset/debt worksheet, to suggest the most equitable division under relevant law.

Divorce and Parent Coaching

AI technology could also be used therapeutically to support family law litigants. Family law practitioners often grapple with the emotional and personal layers of a client’s case despite lacking the relevant therapeutic or mental health experience. AI, along with person-to-person therapy, could help the client separate legal issues from emotions. It could also help a client better understand the need to focus on the legal issues rather than personal ones. This avoids unnecessary legal fees and using court time on non-legal concerns.

Practice Considerations

For now, the family lawyer remains as relevant and needed as ever. Self-representation requires work, and many litigants with means will choose the expert and time saved over fees. AI can assist practitioners with numerous tasks, including case management, mediation, trial analysis, and discovery. It also has the potential, like with clients, to assist with lawyer professional development and mental health. With AI as a tool rather than a replacement, family law practitioners can better use their time to grow their business, efficiently and effectively represent their clients, and dedicate more time to their families and personal endeavors.

Case Law Research

Companies like Westlaw and LexisNexis have already embraced AI technology since ChatGPT-4’s release. Westlaw Precision, for example, uses the GPT large language model and pulls from Westlaw’s database to provide easier access to case law and research.14 Instead of the traditional law school methods of searching with quotes, keywords, and Booleans (like “+” or “-” to expand or limit search terms), a practitioner can simply ask the platform a detailed question. Lexis+ offers a similar service that assists with case research, as well as drafting, summarizing, and analyzing documents, ensuring that documents are “hallucination-free.”15 With AI’s assistance, practitioners can more quickly and accurately obtain the information they need to prepare for trial, advise clients, and ensure they present courts with relevant and up-to-date case law. The time saved with initial research and revising incorrect research allows lawyers to refocus their time on expanding their practice, providing better client counseling, and preparing for trial.

These services can also more easily and readily update the practitioner on new legal trends or case law. Lawyers can type questions like, “Has anything changed with Colorado child support guidelines?” to ensure they are advising clients appropriately or have an accurate understanding of relevant cases.

Discovery and Trial Strategy

Discovery can be one of the most daunting aspects of practice. Receiving, reviewing, and preparing discovery requests demands both time and attention. This process can fall victim to human error when a practitioner overlooks or incorrectly answers a discovery request, or fails to include a request when propounding discovery. Before the rise of ChatGPT, several platforms offered “e-discovery” or “electronic discovery” to streamline the process of identifying, collecting, reviewing, and producing electronic information.16 Family law practitioners often receive large documents—such as bank, credit card, retirement account, or business financial statements—during discovery. Reviewing these statements takes considerable time and effort, especially when identifying a party’s financial activity patterns or understanding the history of growth or depletion of marital assets. Practitioners handling custody matters often receive hundreds of pages of text messages, TalkingParents (a co-parenting communication tool) records, photographs, and audio recordings spanning months to years of interactions between the parties.

Traditional e-discovery tools are often inaccurate and require substantial human input to weed out irrelevant data and analyze documents. Now, new platforms offering AI-backed discovery services—with the potential to minimize human errors—are emerging. BrainSpace is already being used by major law firms (and non-legal firms, such as Deloitte) to accurately analyze and assess relevant data and strategize for trial. These tools can accurately tag and store client documentation or opposing party production, and they will soon be able to consider relevant family law documents when asked questions like, “How much has this asset grown in the past five years?” or “How often does a parent log in to TalkingParents?”

Similarly, AI can help practitioners present their best case to the court if mediation is unsuccessful. Court filings are matters of public record and provide a critical mass of data for AI analysis.17 AI trial software, such as Blue J L&E or Harvey AI, can assist with organizing evidence, identifying exhibits, and connecting pertinent information from exhibits, pleadings, and transcripts with relevant legal elements. AI software can also help draft trial questions, identify important witnesses, and analyze similar cases before a particular judicial officer, which could optimize a case-in-chief. Some of these services are not yet available but are coming soon. Others are already available, including Reveal, which can help identify exhibits that, for example, best exemplify potential intimate partner violence between the parties when asking the court for sole decision making for their client.

Mediation

Family law cases will always require a human touch—meaning that human mediators will remain essential to resolving divorce and custody disputes. AI tools can assist mediators by providing an objective and impartial analysis of the dispute and advising the mediator on key issues and the strengths and weaknesses of each party’s position.18 Services are already available to generate potential solutions and creative ways to solve party disputes over asset allocation or parenting time schedules.

These tools can also, in real time, help mediators find and analyze recent laws and cases that will impact the applicability and enforceability of any mediation agreement. AI assistants with access to court documents and other cases determined by the same or similar judicial officer could also offer insight into how the parties would fare in court versus settling. AI can also be trained and fine-tuned to remove potential implicit biases, making the process more fair to parties involved.

Family law practitioners often face limitations with mediation—the sessions are too short, or clients have limited funds to pay for ongoing settlement talks. Still others may face challenges with client control and unrealistic client expectations, which are hardened by an unsuccessful mediation session. AI services can assist practitioners with better explaining the client’s best day in court versus the benefits (both financial and personal) of settling out of court.

AI services can also assist mediators with administrative tasks, like scheduling and invoicing, to offer more mediation time to family law litigants.

Human mediators will remain crucial to the emotion-filled field of family law. Their skill in managing party (or counsel) anger, frustration, and fear—which may fuel the conflict—is central to the dispute resolution process. Mediators create an environment where participants can express their emotions in constructive ways.19

Judicial Considerations

Colorado dockets are overwhelmed with overworked judicial officers, limited staff, and increasing demand for family law. As mentioned previously, upwards of 80% of family law litigants in the state enter the family law courts unrepresented, demanding additional time from the bench to explain the process and establish reasonable expectations. Many judicial officers enter family law with limited understanding or knowledge of the practice. They often have limited time to familiarize themselves with relevant issues, case law, and practice standards. With the rise of AI technology, judicial officers must consider two things: (1) how to handle AI-generated evidence in their courtrooms and (2) how best to use AI to assist with their decision making and docket management.

Proceedings

The law rarely, if ever, keeps pace with technology.20 From a fact-finder perspective, AI presents several new evidentiary and constitutional considerations. For example, with the rise of AI-powered discovery tools, judges will receive evidence collected and analyzed using algorithms that may violate a party’s due process rights by using unscrupulous or undisclosed means of obtaining information. If exhibits are prepared by AI, judges may receive an argument that a party should be able to impeach the machine at trial if it improperly or inaccurately produced a summary result.21 Judges serving on family law cases, which are often fast-paced with voluminous documents, must be prepared to address complex and nuanced legal issues surrounding AI, such as hearsay, due process, and expert testimony. They are not alone: state and federal courts as a whole are considering changes to their rules of evidence to address AI.

In April 2024, for example, the US Judicial Conference’s Advisory Committee on Evidence Rules invited eight experts to present on AI and machine learning to determine whether changes were needed to the Federal Rules of Evidence.22 Experts such as computer scientists and academics discussed the risks of AI being used to manipulate videos and images and create “deep fakes” that could taint a trial.23 While no definitive rule changes have yet been made, the committee is considering modifying rules involving authentication or identification of evidence in response to AI, such as 901(b).24

Rule 901 is only one of many that will play a pivotal role in AI evidence admissibility. For example, AI sometimes produces misleading, confusing, or cumulative information or documents. Judges will have to consider whether such evidence, even if relevant, should be admitted under Rule 403.25 In Washington, a judge recently barred the admission of an AI-enhanced video, describing AI technology as novel and relying on “opaque methods to represent what the AI model ‘thinks’ should be shown.”26

In family law cases, where the fact finder is also the judge, this presents a particularly challenging situation. How would a judicial officer, unfamiliar with acceptable standards of validity or reliability of AI technology, determine that evidence produced by it is more relevant than it is prejudicial?

Rule 703, which governs expert witnesses and personal knowledge, may complicate authentication of AI-generated evidence. If an AI software generates summary exhibits for a party’s CRCP 16.2 disclosures, for example, a party could not call the data scientist or software developer as an expert witness for authentication purposes because they would not have personal knowledge about the financial information in question. Even if a party were to authenticate their AI evidence under Rule 901(b)(9), which permits authentication by “[e]vidence describing a process or system and showing that it produces an accurate result,” they would have the tremendous task of establishing that the AI system it used was accurate. Not all AI evidence is created equal.27 Some AI systems have been independently tested and shown to be valid and reliable.28 Others have not, when, for example, efforts to obtain information sufficient to test their validity and reliability have been blocked by claims of proprietary information or trade secret.29 Colorado has yet to establish a similar task force, but it would likely adopt amendments that are substantially similar to the federal rules, as it has with amendments to the Colorado Rules of Civil Procedure.30

Because of the technical complexity of AI evidence, family law judges must discuss with the parties, well in advance of the trial, whether they intend to offer AI or similarly technical evidence at trial. If so, the court must include as part of the pretrial process reasonable deadlines for disclosing an intention to introduce such evidence and for challenging its admissibility. These deadlines must be far enough in advance of trial to allow a hearing (which will likely require the testimony of expert witnesses).31 Family law judges should expect to see a wide range of AI-generated evidence, including business valuations conducted by AI-powered software (in lieu of traditional CPA or CFA expert reports) and summaries of financial statements and disclosures created by AI-powered discovery tools. Lawyers, too, must be prepared to present their case for any AI-generated or AI-assisted evidence, including explaining to the court how AI works and how their evidence was generated.

Docket Management and Assisted Decision Making

Courts in some countries have already implemented AI technology. For example, Colombia courts used AI to assist with replacing their docket selection systems.32 In China, AI robots are being used to greet visitors and help guide them to the appropriate location.33 AI could similarly assist family law judges with explaining the legal process, including practice standards, in simple terms that help unrepresented litigants and attorneys alike understand courtroom expectations, deadlines, and procedures. In Colorado court self-help centers, AI could assist self-represented litigants with basic forms, discovery, and pre-trial preparation.

Legal transcription times can also be improved with AI. Legal transcription accounts for 30% of all business transcription and is expected to be worth over $3 billion worldwide by 2029.34 However, delivery can be slow and incredibly costly—especially for parties in family law cases with limited funds. By using AI to assist with accurate and real-time transcription of proceedings, courtrooms can substantially reduce current wait times for transcripts and allow parties quicker access to testimony and their records.

AI also will undoubtedly play a role in drafting routine court documents.35 Much of the work of courts is repetitive and even in high-level courts often tends to follow a standard format.36 Such matters as scheduling or drafting show cause orders can be relegated to AI without grave risk.37 For family law courts, AI will likely assist considerably with trial management orders, notices, minute orders, and other documents currently managed either by the court’s clerk or delegated to counsel on the case.

Unlike humans, AI can sit for more than seven hours a day, doesn’t take eight weeks off for summer, and receives no salary or index-linked pension.[38 The idea of a robot judge is no longer futuristic. Throughout 2019, the Estonian Ministry of Justice and Estonia’s Chief Data Officer Ott Velsberg developed and piloted AI software to hear and decide on small claims disputes of less than €7,000.39 There are drawbacks to computerizing the judicial system, particularly in a field so emotional and deferential as family law. For example, many judgments involve an element of judicial discretion,40 particularly where children are involved. There are also issues translating law into code41 and concerns with how the legislative intent of nuanced statutes like CRS § 14-10-124 (dealing with best interests of the child) could be applied using commands and functions of a computer program.

In some countries, judges are already relying on AI to ensure consistent rulings among courts.42 AI can also be used to ensure that a dispute resolution by a particular court is in line with the results reached by other courts on similar facts and legal issues. For Colorado courts, where most new family law judges have little experience in the field, AI can assist them with their initial decisions to ensure they properly apply relevant case law and statutes on matters such as relocation, UCCJEA jurisdictional questions, and common law marriage claims. With consistent, accurate rulings, appeals are less likely. They will reduce the burden on appellate courts to reconsider a trial court’s order.

Most people still believe that AI judges are less fair than human ones.43 AI is unlikely to replace judges as long as human judges remain central to family law cases; AI can instead assist judges with issuing more consistent, efficient, and accurate decisions.

Conclusion

Practitioners and judges alike have valid fears that their roles will soon be replaced—automated by robots with little feeling or discretion. As AI continues to permeate family law, its dual role as an innovative tool and a complex ethical challenge becomes increasingly apparent. While AI offers significant advantages in case management, client support, and judicial efficiency, it raises questions about preserving the human element in law. To circumvent the unintended consequences of AI’s rise, practitioners must embrace its potential and understand the technology behind it.

The future of AI in family law will be a balancing act, leveraging technology to enhance legal processes while maintaining the empathy and discretion that define human judgment. As we navigate this quiet revolution, ongoing dialogue, ethical considerations, and thoughtful integration of AI are crucial to ensuring that the essence of family law—its remarkably human focus—remains intact. Embracing AI’s potential while recognizing its limitations and our core values will be critical to integrating rather than resisting this new era in our profession’s most emotional and personal practice area.

---

Explore This Issue

September/October 2024 Lawyers’ Announcements

Lawyers’ Announcements

Confronting Strong Emotions

President’s Message

By Catherine Chan

1,260 Sick Shipped Sheep

Historical Perspectives

By Frank Gibbard

Related Topics

Notes