AI in Legal: Disruptive Potential and Practical Realities

We are living through a period of seismic but still unfolding business transformation relating to artificial intelligence (Al), with perspectives shifting regularly from irrational exuberance to scepticism to confusion to alarm, as this week's reaction to DeepSeek attests. 

In the business of law, it is clear that generative Al is poised to revolutionize the legal industry in much the same way spreadsheets transformed quantitative work in the 1980s and 90s. At its core, legal work is fundamentally language-based, making it an ideal candidate for Al-driven innovation. According to a 2024 Gartner report, the rise of Al and legal technology is projected to create a $50 billion market-a headline-grabbing figure that underscores the immense potential of this technological shift. 

As studies have shown, the most significant transformation is likely to occur at the lower end of the skills and experience spectrum. An associate who once required 30 hours to conduct document review might with Al assistance accomplish the task in three hours. 

This newfound efficiency raises important questions-such as, how will legal professionals use the remaining 27 hours? Will they focus on strategic analysis, client engagement or professional development? Moreover, clients are increasingly demanding these efficiencies, with expectations that they will not be billed for time-consuming tasks that Al can handle in a fraction of the time. 

The legal industry's measured approach to Al investment 

Despite eye-catching headlines about Al's potential, recent research by Burford suggests that legal industry investment in Al remains measured. While there is broad agreement that Al can automate routine tasks and improve efficiency, actual financial spend is modest. In a survey of 400 senior lawyers and finance professionals, 89% indicated their organizations would allocate less than 5% of their legal budgets to Al in the coming year. 

I recently hosted a webcast discussion with Jae Um (Six Parsecs), Mike Bryant (Knox Capital Holdings), Jeroen Plink (Legal Technology Hub), and Zach Posner (The LegalTech Fund) on the real growth trajectory of the legal technology market. 

While a clear majority (72%) believe technology will automate routine legal tasks, confidence in its broader impact is more tempered. Only 55% believe that technology could help lower litigation and arbitration costs, and fewer still (48%) believe that Al-driven predictive analytics and algorithms will play a pivotal role in assessing case outcomes and shaping litigation strategies. 

Structural challenges to widespread Al adoption 

This cautious outlook suggests that practical implementation remains a challenge. Factors such as data limitations, regulatory concerns, and the inherent complexity of legal work contribute to this slow adoption. Further, despite its promise, Al faces several structural challenges that hinder its widespread adoption in the legal industry: 

1. Hallucinations and bias: Al models are prone to generating inaccurate or biased outputs, which can lead to costly errors in legal practice. 

2. Privacy and professional privilege: Ensuring the confidentiality of sensitive legal information remains a significant concern. 

3. Lack of broadly available settlement data: Al models rely on vast amounts of high-quality, structured data to generate accurate predictions. The lack of broadly available settlement data creates a challenge, as it restricts the ability of Al to learn from past outcomes, leading to potential biases and incomplete risk assessments. 

In the field of commercial disputes, where Burford operates, one of the most significant challenges to Al adoption is data scarcity. Lacking accurate and representative data, even the best Al will be relatively useless. Commercial disputes are generally lacking in public data. 

Unlike other areas of the law that provide abundant publicly available data, the vast majority of commercial disputes are resolved through private settlements, leaving Al models with access to only a limited and unrepresentative subset of information. Without comprehensive datasets that include critical details such as settlement amount, terms, timing, and other aspects of settlement, Al's predictive capabilities are inherently constrained. 

This lack of transparency in commercial dispute outcomes poses a major structural obstacle to the development of Al in commercial disputes, making it difficult for Al to generate accurate forecasts or provide meaningful strategic insights into potential financial outcomes. As a result, human expertise remains indispensable in interpreting complex commercial legal scenarios and making informed decisions where Al falls short. 

The use of Al in legal finance In legal finance

Al is being leveraged in various ways to enhance processes such as sourcing investments, and to a lesser extent, help to make investment decisions and predicting case outcomes. Legal finance providers like Burford have experimented with Al to identify potential investable cases by combining public data with proprietary insights. This approach has helped surface and pinpoint meritorious investable claims and enable more effective case selection. 

Al is also being used to support underwriting decisions by analyzing extensive case data. However, it remains limited in its ability to replicate the nuanced judgment required in legal finance. Human expertise continues to play a critical role, particularly in assessing the idiosyncratic risks associated with litigation and arbitration financing. 

With proprietary data accumulated over years of case funding, legal finance providers are integrating Al to refine their analytical frameworks and better predict case outcomes. However, Al remains a tool to enhance decision-making rather than replace it. 

The path forward 

The legal industry is on the cusp of a significant shift, with Al promising to drive efficiencies, reduce costs and enhance decision-making capabilities. However, firms must navigate the challenges of integration thoughtfully. A hybrid model that leverages Al's strengths while maintaining human oversight appears to be the most viable approach for the foreseeable future. 

As Al technology evolves and legal professionals become more comfortable with its capabilities, its role will continue to expand. The key to success will be balancing innovation with the appropriate level of human legal oversight and judgment-ensuring that Al enhances, rather than replaces, the critical thinking and judgment that define the legal profession.