Navigating the Trough of Disillusionment: The Evolution of AI in Legal Practice

We find ourselves at a pivotal moment in the adoption of AI in legal practice. A year ago, the buzz surrounding AI was palpable, with forecasts predicting substantial productivity gains and the automation of a significant portion of legal tasks. However, recent research by Goldman Sachs has painted a different picture, highlighting the substantial cost of training and serving large AI models and the lack of meaningful benefits being achieved by users. Rather than viewing this transition from exuberance to the "trough of disillusionment" as a negative indication, Servient interprets it as a sign of healthy market maturation.

The latest Goldman Sachs research, Gen AI: Too Much Spend, Too Little Benefit, includes interviews with various commentators who voice concerns about the current state of AI adoption. Compare this with research published by Goldman Sachs just over a year ago. In Generative AI could raise global GDP by 7% the research predicted that over 40% of all legal tasks will be automated by Gen AI.

What does the stark difference in research outlook tell us about the state of AI in legal tech? The findings illustrate a sentiment that has been growing in the legal tech industry: the time of simple AI wrappers relying on large public language models to perform a single, simple task is coming to an end. The market is demanding more, and rightfully so. 

The trough of disillusionment, as defined by Gartner, is a phase in the hype cycle where expectations are not met, and the technology is criticized. However, it is essential to remember that this is a natural part of the technology adoption lifecycle. It is a healthy sign that we are approaching this phase with legal AI, as it signifies a shift towards increased maturity and a more realistic view of the technology's capabilities and limitations.

This is a natural evolution of technology adoption. As the market becomes more discerning, vendors are forced to differentiate themselves and demonstrate the value they offer beyond the hype. But make no mistake, Generative AI will transform the practice of law. Lawyers who sit on sidelines will do so at their own peril.

Today, the legal profession's focus should be on complex legal workflows delivering real ROI. Successful AI implementations will learn from existing legal workflows to address complex legal tasks. The key to unlocking this potential is the development of agentic workflows. These workflows split complex tasks into manageable steps and incorporate attorney feedback and supervision within the workflow. This approach allows AI to learn from human expertise and adapt to the nuances of legal practice.

It is essential to remember that we are still in the early stages of AI's transformation of legal practice. The internal efforts to build simple wrappers on top of language models have not produced meaningful changes to legal workflows. Software platforms, like Servient's AI Canvas, allow the assembly of agentic workflows to enable AI to deliver on complex legal use cases. The true work of building the underlying technology that will transform legal practice is still in the early stages. 

The trough of disillusionment with legal AI is a healthy sign of a maturing market. It is a call to action for vendors, practitioners, and researchers to focus on delivering real value and to weed out the hype. The future of AI in legal practice is bright, but it requires a commitment to understanding the technology's capabilities and limitations and a focus on delivering real value to the legal industry. Stay the course, keep your head down, and realize that this is a natural maturing of the cycle of technology.