If vendor promotional materials are to be believed, artificial intelligence is the secret sauce cooked into nearly every legal application on the market these days. Familiar legal research and document drafting tools have all been enhanced with AI superpowers, and few legal technology startups dare hit the market without a claim to be supported in some fashion by artificial intelligence.
Everyone is certain that the practice of law will be fundamentally reshaped by artificial intelligence – particularly “generative artificial intelligence,” a technology that can dive into a deep sea of digital information and emerge with enough pearls to make human lawyers question their choice of profession. With good reason too. Financial powerhouse Goldman Sachs believes that artificial intelligence could automate 44% of the work performed by attorneys, an ominous statistic. Fortunately, most experts believe that artificial intelligence will empower – not replace – today’s attorneys.
And while few AI-enhanced legal applications now on the market are specifically directed to solving common professional tasks related to depositions, it’s becoming easier every day to see how artificial intelligence can constructively be applied to preparing for, conducting, and analyzing depositions in a modern litigation practice.
Enter ChatGPT
Artificial intelligence became a hot topic within the legal community this year following the release of a compelling technology called ChatGPT. Many of the new legal applications coming on the market are built on top of OpenAI’s groundbreaking ChatGPT technology. ChatGPT is, in essence, a chatbot that is capable of producing narrative summaries of large amounts of data such as court opinions, pleadings, contracts, and deposition transcripts. ChatGPT’s output is highly customizable. Users can direct ChatGPT to respond in a particular style, language, format and level of detail.
ChatGPT can be customized for legal applications in three main ways. First, by feeding ChatGPT a strict diet of high quality legal information. Second, by developing law-related prompts that extract from ChatGPT information useful for lawyers, in a format that can be adapted to law practice. Third, by including data security protections – technological or contractual, or both – around client confidential information that might be processed by ChatGPT.
In the right hands, ChatGPT produces output that looks like it might have been written by a lawyer.
ChatGPT, along with other generative artificial intelligence technologies, is a technology advancement over former data analysis tools because of its ability to glean legal concepts and insights from datasets rather than mere character string matches detected by search technologies. The shift from detecting similar words to detecting similar principles, and expressing them as a human might, moved computers much closer to the lawyer’s domain.
Applying Artificial Intelligence to Depositions
Artificial intelligence promises to be useful for a number of common deposition-related tasks:
- summarizing documents to highlight or unearth key issues
- summarizing materials obtained during pretrial discovery
- drafting a questioner’s outline for use during the deposition
- researching local laws applicable to the deposition
- performing background checks on deposition witnesses
- conducting sentiment analyses of witnesses, based on a written transcript, audio or video recording
This list isn’t exhaustive, of course. The ingenuity and deposition experience of modern litigators will no doubt discover additional uses for artificial intelligence in deposition practice.
“Everyone is certain that the practice of law will be fundamentally reshaped by artificial intelligence – particularly “generative artificial intelligence,” a technology that can dive into a deep sea of digital information and emerge with enough pearls to make human lawyers question their career choice.”
Looking to the technologies currently available to litigators, the following legal applications all claim to be useful specifically for deposition practice. We’re not endorsing any of these technologies, or representing that they live up to their marketing claims. They’re mentioned here to give a glimpse of how technology might be applied in new ways to deposition practice in the not-too-distant future.
Casetext CoCounsel promises to boil a trove of documents down to just the most important information, draft a deposition outline, suggest fruitful areas of inquiry, spot weaknesses in theories of recovery, and provide a summary of the local laws applicable to the deposition.
Everlaw Story Builder, a post-deposition case analysis tool, highlights and collects important case documents and deposition testimony into a single location, using them to construct timelines and a compelling case narrative for use at trial.
Similar to Story Builder, DISCO Case Builder is a technology that allows users to organize and draw insights from deposition testimony.
Sullivan & Cromwell, a global law firm based in New York, is reportedly developing an AI Deposition Assistant that can create deposition questions and flag unusual witness “body language or discomfort in real time.”
Apart from these deposition-specific tools, there are a number of AI-enhanced offerings on the market that could be of assistance to litigators preparing for depositions.
Legalyze.ai, for example, creates summaries and timelines based on case documents, all of which may be useful for deposition preparation. Humati.ai offers a similar AI-enhanced document review and summarization tool. Leah CoPilot from ContractPodAi promises the capability to summarize legal documents and “discover connections that might elude a human.”
Reveal AI supplies litigators with pre-trained AI models on legal subject matter and “sentiment analysis” tools that detect the tone and intent of communications, and whether the witness is discomfortable or testifying dishonestly. Reveal also provides technology that provides detailed background checks on deposition witnesses, which may be useful for developing lines of questioning.
Traditional legal research giants Thomson Reuters (Westlaw Precision AI) and LexisNexis (Lexis+ AI) have come out with their own AI-enhanced products that, while not specifically directed to deposition practice, have the potential to significantly streamline common deposition preparation tasks. Unlike plain vanilla ChatGPT offerings, which are trained on freely available Internet data, both Precision AI and Lexis+ AI draw their insights from large databases of proprietary legal information.
Finally, there’s Harvey.ai, a not-yet-available ChatGPT implementation specifically customized for legal applications that emphasizes data security. Earlier this year, Harvey.ai announced a partnership with international law firm Allen & Overy, where Harvey is being road-tested by its over 3,500 lawyers.
We haven’t yet reached the point where any of these technologies, however exciting they may be, will replace or even approximate the services of a skilled litigator. However, it’s undeniable that the same skilled litigator, armed with artificial intelligence, will be more efficient, more valuable to clients, and perhaps more compelling in court than an equally skilled litigator without technology assistance.
In fact, the choice whether or not to use artificial intelligence in law practice may not even be a choice at all. Although the lawyer’s ethical duty of technology competence does not require the use of technology, it does, at a minimum, require lawyers to be knowledgeable about relevant technologies and to engage with them. Lawyers should consider, at a minimum, whether there is a technology that could be appropriately applied to the case matter at hand and to weigh both the risks and benefits of doing so.
A thoughtful summary of ethical issues raised by artificial intelligence, Practical Guidance for the Use of Generative Artificial Intelligence in the Practice of Law, was published Nov. 16 by the State Bar of California. The California guidance touches upon each of the ethical obligations (there are over a dozen) in the lawyers’ professional code that are implicated by the use of generative artificial intelligence technologies. Leading concerns mentioned are the perils of over-reliance on the outputs of generative AI, the security risks to client confidential information passed as inputs to generative AI, the need to disclose AI use to clients, and considerations for billing (or not billing) clients for expenditures on AI technologies used during the representation.
