A young attorney is a guest at a dinner party hosted by one of his firm’s partners. As the evening proceeds, his colleagues begin to take turns telling their favorite lawyer joke, sparking much self-effacing laughter.
Not wanting to come up short, the associate loads ChatGPT on his smartphone and enters, “Tell me a lawyer joke.”
When heads finally turn his way, he takes one last look at his phone to double-check the results and then asks the room, “Why don’t lawyers go to the beach?”
After waiting for a few shoulders to shrug, he says, “Because cats keep trying to bury them in the sand!”
Instead of laughter, the other guests meet the young lawyer’s gag with scoffs and soured faces. Within seconds, his complexion is cherry red and he’s wishing he could sink through the floor – or that he’d never heard of ChatGPT.
Although this scenario is fictitious, it’s been within the realm of possibility since the launch of ChatGPT, an artificial intelligence chatbot developed by OpenAI, in November. Also, while the dinner party is a fabrication, the joke isn’t; the AI actually produced it in response to a query for a lawyer joke.
One can imagine the very real New York City attorney Steven Schwartz felt like the associate in the story when he learned the brief he submitted for the case of Mata v. Avianca in federal court was no laughing matter, either, but was riddled with citations of cases that didn’t exist.
Schwartz used ChatGPT to supplement his findings in a personal injury case in which a client was suing an airline after a serving cart battered his knee during a flight, Mashable reports. ChatGPT obliged, providing a list of cases Schwartz must have thought sounded reasonable: Petersen v. Iran Air, Martinez v. Delta Airlines, Miller v. United Airlines and more.
The problem, says Chattanooga attorney T.J. Gentle, is that OpenAI has not yet programmed ChatGPT to be factual; rather, at this stage in its development, the chatbot is designed to be merely convincing.
“Generative AI like ChatGPT seems to be able to reason, and by seeming to be able to reason, it instills trust in us and creates an opportunity for us to be misled,” Gentle says. “As humans, we learn through mimicry, particularly with respect to language. Generative AIs are pure mimicry, but they have been trained on so much underlying data that the mimicry is uncanny.”
With this in mind, Gentle says people who are relying on ChatGPT to make complex decisions are doing so at their own peril – as Schwartz has also expressed in apologetic statements.
“Sometimes, it will make stuff up,” Gentle continues. “It’s not doing it to be nefarious; it’s doing it because there are quirks in the underlying data.”
Thus, as the world attempts to grasp the implications of AI, the first major story to emerge from the legal arena is not one of triumph, in which an AI devised the winning strategy in a contentious trial, but that a lawyer allegedly failed to do his due diligence and, in the process, might have thrown his 30-year legal career in a Manhattan dumpster to boot.
Gentle, a mergers and acquisitions attorney with Miller & Martin, understands a thing or two about AI. While working for Smart Furniture from 2005-2015, he contributed to an effort at the Chattanooga company to use rudimentary AI to improve the recommendations its website made to customers, as well as to improve its web-based marketing. In the process, AI came to fascinate him, he says.
Gentle’s research led him to regard AI with both awe and trepidation. On the one hand, he believes it has the potential to do great good; on the other, it could do great harm. Either way, it will trigger dramatic societal shifts, he says.
“AI will be incredibly transformative. I believe it’s the most significant technological development in my lifetime.”
As points of comparison, the 46-year-old Gentle has also witnessed the emergence of smartphones, 3D printing and the Human Genome Project. However, he says AI will eclipse their impact as it touches every aspect of human existence – including his profession.
AI v. The Cotton Gin
Humans have been leery of AI since the beginning of the computer age. As early as 1968, filmmaker Stanley Kubrick’s science fiction classic, “2001: A Space Odyssey,” featured a sentient computer, HAL 9000, as its antagonist. When a malfunctioning HAL learns the astronauts on a manned mission to Jupiter intend to shut it down, it attempts to kill them.
Through “2001,” Kubrick explored the early existential questions to which the appearance of computers gave rise. However, Gentle says AI won’t do away with attorneys but rather make them more efficient.
For example, he posits, AI is poised to accelerate the process of preparing legal documents and drafting agreements.
“If you’re a lawyer who makes money doing commodity work or filling in the blanks on forms, AI will render that work obsolete. That work will be gone because inexpensive or free AI-based software will be trained on all the forms in the public domain, and that software will produce simple legal documents more efficiently than any attorney.”
Attorney Billy Eiselstein, a member of Miller & Martin’s Litigation Practice Group and co-chair of the firm’s Privacy and Data Security Group, says AI is already expediting the laborious process of document review in litigation and will only become better at performing this task as its algorithms evolve.
“We already have an AI tool that finds documents based on searches for documents that address specific issues. Instead of having a someone comb through a stack of documents one at a time, the AI can find the relevant documents for you. As AI becomes more and more powerful, that will continue to be a trend in the industry.”
While the prospect of spending less time on legal grunt work might sound like a reason to celebrate, it introduces the problem of Eli Whitney’s cotton gin – likely the most significant technological advancement of 1794. While Whitney’s machine sped the process of removing seeds from cotton – and made the U.S a leading exporter of the commodity – it also led to the growth of enslaved labor.
Like the cotton gin, AI will have what some might argue will be a negative impact on the labor force, Gentle says.
“Industry experts estimate AI can automate 20 to 40% of legal tasks, most of which are administrative. However, new technologies have always impacted the legal profession. When I started at Miller & Martin in 2000, one secretary would work for two attorneys; today, one secretary works for several attorneys because the tools we use have streamlined our work and our attorneys and secretaries have become better at using them. So, there are fewer openings for paralegals and secretaries than there used to be, and I expect that will continue to be the case as AI proliferates.”
While Eiselstein agrees AI will reduce the need for people to perform ground level tasks, he says he believes artificial intelligence won’t necessarily lead to a loss of jobs but instead change the nature of the work employees do.
“We’re going to need people who can interface with the AI and verify the results, so their skillset might change or evolve.”
Gentle says he’s looking forward to AI liberating him from the more mundane tasks of his practice and freeing him to focus solely on its more cerebral endeavors.
“The thought of using AI to do the parts of my work that don’t require me to use my brain is exciting. In two to three years, AI will be able to produce first drafts of an agreement that are better than what exist now. I’ll be able to activate the microphone on my computer and describe what I need in an agreement, and the AI will scour every document that’s in the public domain, in addition to every document I’ve produced, and create something good.”
AI v. Lawyers
As Gentle anticipates putting his mind to better use, he says he’s confident he’ll still have a job that requires him to use it. While the world at large wrestles with fears that AI will someday either rule or eradicate humankind, he says he feels safe in predicting it’s not yet capable of replacing the lawyer as an essential component in business and litigation, partly because clients still want a thinking, reasoning human being at their side, whether they’re seeking to acquire a business or defend themselves against a murder charge.
“AI will have a difficult time replicating – or automating – what a client really needs. For example, if my client wants to acquire a new technology in an acquisition, part of my job might be to determine if the technology has any legally protectible IP rights, along with whether my client will be able to rely on those protections after the deal closes.
“Helping my client to understand what he’s acquiring, as well as the legal risks and limitations associated with that technology, involves a significant level of analysis and reasoning. A large portion of that reasoning might be based on legal training, but an equally important portion is based on my ability to communicate with the client and to make rational inferences from the information available. At this point, AI technologies cannot provide this level of reasoning or analysis.”
Ultimately, Gentle continues, when a client is about to buy a business for $50 million, they want to look into his eyes – not into HAL’s unblinking red eye – and consider whether or not they trust that he has their best interests in mind.
“That will always be there,” Gentle insists.
As evidence that a lawyer can tell a better joke than ChatGPT, Gentle says another reason attorneys can breathe easy for now is the possibility that AI could create more problems than the current batch of lawyers can solve – and thereby create a need for more of them.
“As people in Silicon Valley create these life-altering technologies, they ignore the limitations in society; they say, ‘I can create a tool that can do amazing things,’ but they don’t think about the status quo, or the rights of the people who will be impacted,” Gentle speculates. “However, the use of AI is going to break a lot of things.”
Gentle says the creation of content by ChatGPT can serve as an example of how the use of generative AIs raises new legal questions and will likely change not just the practice of law but also the law itself. “There are already questions about authorship and ownership of the underlying content, so our copyright laws will change,” he reasons.
(It’s likely no one will claim to have written the Gentle profile ChatGPT generated June 24, however. As if to prove Gentle’s point that the AI will make up content when it finds none, the article inaccurately reports that a childhood accident left him paralyzed from the waist down and that he later became a wheelchair basketball star who inspires others “one step at a time.”)
When Gentle speaks with people about the larger challenges AI poses to society, such as the repercussions of using artificial intelligence to display social media content that continually engages the user – a technology he argues creates an echo chamber that reinforces only one point of view – he recommends they step cautiously into the future and take measures to ensure the technology is used in ways that benefit rather than harm humankind.
With Schwartz’s folly fresh in their minds, Gentle and Eiselstein recommend those in the legal profession use the same measure of caution. Eiselstein also says if emerging AI tools for legal research, such as Lexis+ AI and Westlaw Precision, offer the safeguards the practice of law demands, he believes attorneys will put them to good use.
Even then, Miller & Martin would do its due diligence, he adds.
“We’d test it to see how well it appears to work, and we’d require safeguards against being blindly reliant on it and for making sure we’re not running afoul of any court rules.
“We also might have clients who would want to weigh in on whether or not it would be acceptable for our attorneys to use those tools. Since they’d be paying us to do the work, we’d need to be transparent with them about how we’re doing it. We might also have clients who, if there are reliably confidential tools, might encourage their use for efficiency.”
As AI tools propagate in the legal profession, law firms in Chattanooga are preparing for their arrival. The leadership at Husch Blackwell, for example, is evaluating “legal industry specific generative AI solutions” they believe will add value and increase the efficiency and speed of legal service delivery.
Additionally, the firm in the process of hiring an artificial intelligence solution strategist to assist in exploring the capabilities of these technologies after those evaluations are complete.
“We’re taking a measured approach to ensure the solutions are secure and meet our legal and ethical obligations,” says Michael Alston, the managing partner of Husch Blackwell’s Chattanooga office.
This approach is surely preferrable to copying and pasting case citations into a brief, or asking ChatGPT for a joke. Then again, if the imaginary young attorney at the pretend dinner party had tried again, ChatGPT might have generated a better punchline, as it did in real life:
“Why don’t lawyers go to the beach?” it asked again. “Because even the sand can’t stand their briefs!”