AI in the Legal Field
Since the inception of generative models, artificial intelligence (AI) has become a de facto concern for public affairs, particularly in the legal field. Experts and journalists, whether they consider AI an opportunity or a threat, expound on its immediate and profound impact on work processes. Professionals who master the new tool, and understand its potential and limitations, are widely seen as having an edge over those who resist innovation. So how should legal professionals put aside the hyperbolic enthusiasm, doomsaying, and fear to fruitfully use AI?
For many years, the legal field has adapted to and greatly benefited from technological change. Access to comprehensive legal databases and online jurisprudence, for example, has significantly streamlined legal research and practice. Generative AI, however, has unleashed concern that it may replace legal professionals. A balanced perspective is needed.
On the one hand, AI can support and augment the work of legal experts by enhancing efficiency and accuracy in tasks such as research, document review, document drafting, and, potentially, predictive analysis. On the other hand, some aspects of legal reasoning, at least for now, cannot be replicated by natural language processing technology. The ability to delineate between AI’s potential and limitations is the foundation for a level-headed approach and a demystification of the technology. It would also empower legal professionals to take full advantage of AI.
The technology’s potential includes quickly sifting through thousands of legal documents, identifying keywords and relevant case law. AI can use templates to draft new contracts and analyze them to identify clauses that deviate from standard practices, noting areas of potential exposure that can be the subject of negotiations. Technologies such as natural language processing enable AI to categorize information, which also increases efficiency, and AI-powered chatbots can provide first-level information on common legal queries. Using historical data and machine learning, AI may one day even predict the outcomes of legal cases, which can help lawyers assess risk, formulate strategy, and consider settlement.
The technology’s limitations and shortcomings for the legal field are more numerous. First-generation chatbot models are unable to process lengthy and nuanced prompts, therefore sizable documents cannot be included in a question. Retrieval Augmented Generation (RAG) could be used to overcome this and extract information from documents, thereby feeding a smaller amount of data to a prompt. But the benefits of this option are limited as analyzing complex documents requires interaction among them. AI-generated knowledge graphs can overcome some RAG shortcomings, but this option is still in its infancy.
Still, it may be human “soft skills” that are the most prominent obstacles to AI’s replacing the individual mind. The technology cannot exercise nuanced ethical judgment and discretion. Nor can it consider a vast array of signs and written and unwritten prompts, all of which are cornerstone attributes of legal professionals. The human mind can assess not only the legal merits of a case but also its ethics, risks, and broader social impact in addition to client intentions, emotions, and preferences. The individual also offers the empathy, reassurance, and care that clients often need. This profoundly human touch is essential to navigating complex legal and socio-political landscapes, social values, historical definitions of words, evolving legal standards, and political contexts. And AI certainly cannot argue for clients in court, where persuasive communication, charisma, strategic thinking, and constant adaptation are crucial.
AI systems can process only what they are fed. They operate within the parameters of their programming and the data on which they have been trained. They struggle with novel or unprecedented legal scenarios that require creative problem-solving and adaptation, areas in which legal professionals excel. AI systems lack the ability to harness the deep, experiential knowledge that seasoned legal experts acquire over years of practice and apply to deciphering subtle nuances and foreseeing potential pitfalls in legal documents or strategies.
Finally, AI underperforms on addressing large operations, such as defining the strategy for a proceeding (or multiple proceedings linked between them) or coordinating the multiple aspects of a merger or acquisition.
Legal professionals strive to be unburdened by tedious tasks, and AI can help achieve that goal. But humans must retain the responsibility for being creative. In this sense, people and machines occupy their own spaces.