Discovery may alter that imbalance. But attorneys, unlike the three-dimensional chess players with whom they sometimes are compared, must account for the telling effect of a fourth dimension: time. Especially in trust disputes, clients may be elderly and in need of testamentary distributions for medical or other urgent needs. Investment opportunities may be fleeting if finances remain uncertain. Bequests may be insufficient to warrant substantial investment in litigation. If parties are intent on resolution rather than retribution, attorneys may be constrained to seek mediation before the costs of litigation exceed client capacity.
Confronted with such dilemmas, attorneys will soon be able to use artificial intelligence (AI) platforms as tools for mediation. Wisdom may be unnecessary for the achievement of settlement on acceptable terms. Nor would knowledge, although perhaps desirable, be required. But information would be essential. AI may serve that purpose, providing an economical tool for the retrieval and formulation of critical data.
AI Platforms
AI platforms that may support litigation are beyond the rudimentary design phase. Two general types are involved: (i) rule-based and (ii) learning-based. From programmed instructions, next-level decisions are proposed: If X occurs, Y follows; or instead, if A is offered, B is posed as a viable response. With ever-expanding databases, such guidance may be constantly refined.
Building upon these systems, two AI technologies are emerging: large language models (LLMs) and generative AI models. LLMs are conceived to handle language-related functions. Drawing from myriad sources, LLM programs can produce text that may be coherent and appropriate to the context of the query. Generative AI models are designed to produce creative artistic, musical, and literary content. Able to discern patterns and characteristics from troves of data, generative AI programs can produce unique images, imaginative sound combinations, and seemingly understandable writings. The sophistication of both LLM-generated text and generative AI creations will increase without seeming limitation.
Eventually, a combined LLM and generative AI platform will be developed for mediations. Attorneys will present their evolving positions to a virtual mediator, an electronic creation of algorithms. The mediator will convey offers and counteroffers between the attorneys. Depending upon the sophistication of its guiding program, the mediator may offer analyses of and recommendations about the proposals based on information retrieved from a database of comparable cases.
This article, however, is not intended as an exploration of the demise of the human mediator. Such advanced technology is noted merely as the foundation for discussion of what will emerge sooner for attorneys. Less robust programs will complement the work of attorneys in preparation for mediations.
As noted earlier, attorneys may occasionally be compelled to commence mediations despite discomfort with the information and knowledge they have. Two examples illustrate such lacunae in trust litigation: (i) allocation of assets between subtrusts and (ii) accountings of profits and expenditures. AI may fill these voids.
Subtrust Allocations
As a significant facet of their structure, family trusts may provide for the allocation of assets to subtrusts upon the death of the settlor. One purpose served by this structure is to minimize inheritance taxes that may be owed upon the death of the settlor. Another reason is to ensure, if feasible, that assets will be available for inheritance by remainder beneficiaries after the death of the trust’s principal beneficiaries, usually the settlor and spouse.
Not infrequently, successor trustees neglect to make the requisite allocations until compelled to do so by legal action. When such late allocations are undertaken, disputes may arise over the appropriate valuation of the subtrusts. In elemental terms, the argument is that, if allocations had been made properly at the time of the settlor’s death, the subtrust subject to funding should be valued at an amount higher than proposed by the successor trustee.
The calculation of such valuations, however, may be complex.
As a relatively simple example, suppose that a family trust holds stocks A and B at the time of the settlor’s death. Over time, A and B may increase or decrease in value, each at different rates. Each also may have dividends and splits in shares. Suppose further that the family trust specifies an equal distribution of assets to be made at the time of the settlor’s death between two subtrusts X and Y.
Under one commonly proposed scenario, A and B may be equally allocated between the subtrusts at the time of settlement. But such a post hoc distribution may not be fair to the remainder beneficiaries, who, for purposes of this illustration, are the recipients of subtrust Y. Most frequently, family trusts will specify that the proceeds from subtrust X, both interest and principal, may be used for the health and welfare of the principal beneficiaries and that interest and principal from subtrust Y may be invaded only under limited circumstances. Subtrusts X and Y thus are likely to have different values within months after the settlor’s death.
Attorneys can use basic spreadsheet formulas to chart the financial condition of subtrusts X and Y over time. The process is time-consuming and laborious, dependent upon accurate initial valuations, stock histories, and expense records. Attorneys may be relegated to positing different scenarios untethered by factually accurate calculations and projections.
AI will be able to support attorneys with these kinds of assessments. Programs can chart different ways that initial allocations could have been made, with corresponding balance projections for each. Programs can also check the validity of any projections offered by opposing attorneys. The savings to attorneys in terms of both time and money if they are not forced to perform this function manually should be substantial.
Trust Accountings
Although timely requested, trust accountings are not always timely filed. Accountings, especially those spanning years, are costly to prepare. Financial records must be retrieved and reviewed. Lines of data, perhaps numbering into the thousands, must be organized and entered.
Delays in the production of accountings may be exacerbated because successor trustees may be incapable of maintaining records and balancing even a rudimentary two-column ledger and may wish to withhold information indicative of misappropriation of funds. Moreover, records furnished through discovery may be incomplete. Subpoenaed bank, financial management, and credit card statements will yield only numbers. If clients cannot wait for these matters to be fully resolved, attorneys, as previously noted, may be compelled to seek relief through mediation under less-than-ideal conditions.
AI may soon enable attorneys to formulate accounting data in multiple ways. Entries may be categorized by group, and groups may be sorted in accounting format. Analyses of patterns of expenditures may be feasible. AI may constitute a tool to enable attorneys to acquire information that otherwise would not be accessible within imposed time constraints.
Attorneys currently perform these functions manually, but only if they have time and resources. AI will provide tools to perform them expeditiously and economically.
AI also will be effective in addressing other functions that similarly involve processing large sets of data. At a casual glance, one such operation in the trust context entails asset valuations. For real estate in particular, valuations are dependent upon comparisons of similar properties in the same location over a specified period. The subjective quality of assessments of similarity, sameness, and time may be enhanced or diminished depending upon the breadth of the data used. Another operation, one that is encountered in not just trust, but all litigation, involves transcripts, briefs, and other documents. Volumes of testimony, arguments, and related information often must be summarized, indexed, and cross-referenced. Such data may be useful only if subject to efficient retrieval.
The application of AI will be limited only by the imagination and creativity of attorneys. Attorneys will become increasingly dependent upon AI to handle volumes of information. The objective presumably will be the knowing use of that information. Whether wisdom will follow may then be left for later determination.
