This article may be biased – but so is ChatGPT.
With the public launch of ChatGPT and other large language models (“LLMs”), many in the legal industry started wondering how the new technology will change their lives and whether their jobs will be eliminated. Like many things surrounded by hype, AI spurs a wide range of emotions: from terror and caution to enthusiasm and hope. In this article, we will provide a sober point-counterpoint discussion of why AI will not replace lawyers but will revolutionize the profession and the business of law, yielding better outcomes for clients.
1. AI Passed the Bar: Should You Sack Your Lawyer Now?
Many AI skeptics were taken by surprise when Stanford Center of Legal Informatics published research about GPT-4 passing the bar exam in the 90th percentile this past April, proving that the legal profession is not immune to the AI revolution.
GPT-4 was trained on a vast corpus of data found online or licensed by OpenAI, including (one would suppose) loads of bar exam prep and other legal content. Given that the bar exam questions are somewhat standard and revolve around the same broad and not specialized subject matter (and, in one of the components, all applicable law and facts are provided), it is impressive but unsurprising that GPT-4 aced it.
Should we now think that AI will replace lawyers? Not exactly. All in all, passing the bar is just the first step to becoming a good attorney. Developing real-life experience and a deep specialized expertise within a larger perspective, keeping abreast of the constantly changing legal landscape, being an effective communicator and advisor to clients, are just some examples of things that human professionals are still unbeatable at. As with AI and statistical modelling in professional sports, technology can enhance some aspects of the game but would unlikely ever replace the human experience and emotion.
The most important thing to understand about large language models is that, despite the splendor with which they generate relevant information, they do not understand the meaning of anything, let alone comprehend how to navigate the complex logic of legal systems as applied to the bigger picture of a client’s particular circumstances. An obvious consequence of this is that LLMs often “hallucinate” –fabricate inaccurate information that sounds confident. LLMs can sometimes declare living people dead, make up case precedents, and generate defamatory claims. So, anyone using AI outputs in scenarios requiring high accuracy (like legal matters) should be able to verify their accuracy.
Another thing to consider is “garbage in = garbage out”: to produce good results, computer systems in general (and LLMs in particular) need to be fed good prompts from someone who understands the issues, the goals of the project, and even the personality of the subject. There are other considerations for responsible AI use, but they exceed the scope of this article.
Large language models are good at anticipating relevant information, much like predictive text in messaging apps. However, the greatest challenge that AI does not seem to overcome is understanding the complex logic pervading the legal corpus and applying what is relevant to the complicated individual circumstances of each unique client or matter.
The law (especially in this country, given it is a case law system and a federation) is a decentralized system with a complex logic that needs to be understood to be successfully navigated. So far, it is the prerogative of human lawyers who undergo extensive training, have years of practical real-world experience, and are professionally liable for their product. This is why we caution clients against the mindless automation of their legal and compliance needs. The financial resources saved in the process will be well exceeded by the clean-up costs of addressing the compliance, contractual, and reputational damage resulting from the reliance on bad advice. And blind reliance on AI-generated output is a recipe for bad advice.
2. If AI Won’t Replace Lawyers, How Will It Affect the Clients?
The legal market has largely been operating under one or a combination of two models: (1) the billable hour; and (2) alternative fee arrangements (AFAs) that focus on the value provided to the client and involve some degree of client control over the legal services bill through fixed fees, fee caps, and the like.
The integration of AI into the practice of law is set to transform traditional business models. If used well, AI can greatly increase lawyers’ efficiency, accelerating routine and time-consuming tasks like document review, contract analysis, and legal research. This will save lawyers time and enable them to focus on higher-level strategic thinking and complex legal matters, as well as provide time to take on more matters. As a result, legal professionals will be able to allocate their time and expertise more effectively, leading to enhanced client service and client satisfaction.
One of the most significant advantages of employing AI in the legal profession is its potential to reduce costs for firms and their clients. Traditionally, clients bore the burden of extensive billable hours spent on repetitive tasks, gathering, and organizing information. With the implementation of AI technologies, lawyers can expedite these processes, significantly reducing the time and resources required.
Despite the market’s demand for AFAs, most big law firms have so far been reluctant to make a shift from the billable hour model. It is interesting to see how the dynamics will change with the widespread adoption of AI productivity tools (including LLMs) by more agile specialized law firms (like, for instance, Marashlian & Donahue, The CommLaw Group).
3. A Challenge to Be Addressed: Training Future Generations of Lawyers and Developing Subject Matter Expertise
There have been concerns in the legal community about how junior attorneys and legal professionals will develop their subject matter expertise in the coming era of AI-assisted legal practice. After all, AI is now being tasked with something that junior professionals used to do, like document review and creation of first drafts. Another thing we mentioned earlier is that, to receive high-quality output from AI, someone with subject matter expertise needs to produce high-quality prompts first. How will future professionals obtain expertise to be good prompters? How is AI going to affect the quality of our lawyers and legal professionals down the road and what do we do about it?
Throughout history, technology has rendered some skills obsolete. We no longer have to rewrite documents by hand, use a typewriter, or go to legal libraries to research case law. Of course, AI technology is much more sophisticated than most technology we encountered in the past. However, we see the same overarching logic: lawyers and legal professionals did not lose their quality but developed new skills and started focusing on other, less mundane, tasks. Moreover, the legal landscape has become more sophisticated and specialized over the years, so lawyers and legal professionals did not catch their breath and started doing new kinds of work.
Law firms, law schools (and, perhaps, even secondary schools) will need to step in, teaching about effective AI use. One thing is certain: in the age of deepfakes and AI hallucinations, critical thinking skills will be as important as ever.
Conclusion
The legal industry must recognize AI as a valuable productivity tool, understanding both its benefits and limitations. Despite the ever-growing Fear of Missing Out fueled by the media, it is important to resist the urge to dive into AI adoption headfirst. There will be a learning curve to the industry’s adjustment to the new AI era. However, by embracing AI, innovative and agile law firms can pave the way in developing efficient processes and fostering the growth of young talent’s expertise. AI is here to stay, and we at Marashlian & Donahue, The CommLaw Group are enthusiastic about its responsible use for the benefit of delivering exceptional value to our clients.
