Artificial intelligence is the replication of human intelligence by a machine. Generative artificial intelligence takes this one step further by leveraging the power of computers, collected data and machines to mimic the problem-solving and decision-making capabilities of the human mind. This leap in machine ability can be a powerful tool to streamline work product. Although advisors might be wise to integrate AI into their practices, estate-planning and family office professionals should remember that a machine can’t replace the human relationships that we build with our clients.
Consider the Following Scenario
Your long-time client calls to share that the mental condition of her mother has rapidly declined. Your client would like to review her mother’s existing estate-planning documents to understand what to expect when the inevitable occurs. Additionally, your client wants to determine the size of her mother’s estate and her authority to manage her mother’s affairs at this stage in her life.
How Can AI Be Leveraged?
That call — a grieving daughter reaching out to discuss a delicate and emotional life occurrence — can’t (and, frankly, shouldn’t) be supplanted by AI. However, GenAI tools exist that might help an advisor more efficiently respond to the client’s needs.
GenAI can potentially review and summarize complex estate-planning documents and asset compilations. In our scenario, you might upload the ailing mother’s financial and estate-planning documents to an AI server and use it as a quasi-search engine. You could ask the machine a simple question, such as “What is the net worth of my client’s mother?” You can ask a more complex question, such as “Would my client’s mother benefit from setting up a private foundation or sale to an intentionally defective grantor trust?” Though fleshing out the answers to these questions might save significant time (especially if the explanations are passed on to the client, as discussed later), advisors should review the answers closely for accuracy.
Critically, advisors must pay special attention to the data privacy considerations of providing confidential client information to AI platforms. Studies indicate that while “the availability of users’ private data enables AI systems to perform better, there are also considerable risks associated with this data collection. One of the main issues is the usage of data for non-intended purposes. Users are often unaware how their data will be processed, used and even sold.”
Even if a software agreement prohibits the sale or disbursement of user information, uploaded data might be used in other ways. There’s little research on the reliability of data privacy assurances made by AI companies. Moreover, state and federal regulation hasn’t caught up to the burgeoning industry. In our scenario, let’s assume that our client’s mother’s trust includes distinct and artfully drafted provisions. If this language were to be integrated into the software’s data bank, identifiable and individual estate-planning provisions might be compromised. The risks of using AI in an emerging data privacy landscape shouldn’t be overlooked.
GenAI tools may be particularly helpful for document drafting. Large Language Models can provide an initial draft of standard estate-planning documents. In our scenario, the client might have her mother named as a fiduciary in her own estate-planning documents. It might prove wise to update those designations or review the current successor structure. LLMs may one day have the capacity to produce an initial draft of basic documents, like a power of attorney or will.
Of course, advisors using AI drafting capabilities shouldn’t treat any AI-produced document as a final draft. LLMs and any AI-created material should be thoroughly reviewed and edited prior to use. As Bloomberg has reported, “poor accuracy is one of several issues making law firms wary of [G]enerative AI. It’s already led a New York federal court to threaten sanctions against attorneys who submitted a legal brief using ChatGPT-generated research that cited non-existent case law. Judges elsewhere are beginning to install guardrails on the use of the technology.” In fact, a Texas judge has issued an order requiring all lawyers to certify that their filings haven’t been drafted by AI. If any AI-generated language was used, the order requires it to be checked for accuracy “by a human being.”
As the technology develops, the estate-planning drafts produced by AI are likely to increase in quality and accuracy. Nonetheless, advisors should review documents with a keen eye. Estate planning is more of an art than a science. Detailed knowledge of a specific client’s circumstances inform how documents ought to be drafted. So, while AI-produced materials might be accurate, they should be refined and tailored to the client’s needs and preferences.
A core skill of an advisor is the ability to explain sophisticated concepts in a way that can be understood by the client. AI may be used to translate legal jargon into a more easily digestible form. For example, you might ask a GenAI software to “explain per stirpes to a 10-year-old” or to “explain the concept of delayed revocation in a health care power of attorney in Illinois.” The software will provide a simplified explanation of exactly that.
Using AI to simplify complex tax and estate-planning concepts might be the most powerful resource available to advisors because it doesn’t require the input of any private or sensitive client information.
GenAI software is a developing tool that’s likely to contribute to the productivity and efficiency of family office and estate-planning professionals. However, clients and advisors alike should remember that some conversations and tasks require a discrete and human approach. By maintaining an open mind to the possibilities of GenAI technology and remaining attuned to the security and sensitivity of our practice, advisors can leverage AI in a myriad of ways.
Additional research and writing from Zoe Belford, a 2023 summer associate in ArentFox Schiff's Chicago office and a law student at University of Chicago.