ANALYSIS: AI Fever in Legal Industry Cools Off, Survey Suggests

Artificial intelligence was quite the hot topic for the legal industry this summer, but is the interest dropping with the temperatures? The legal industry appears to have plateaued this fall with generative AI usage after a summer of full steam ahead, according to recent data from Bloomberg Law.

Bloomberg Law’s most recent State of Practice Survey asked more than 450 law firm and in-house attorneys about their familiarity with generative AI and how their organizations have responded to the booming of the technology.

Fifty-six percent of attorneys reported this fall that they’ve used generative AI, which is a slight decrease from the usage reported in the summer (60%), but still a greater percentage than respondents in the spring (37%). Usage in a professional setting remained relatively consistent, decreasing only by 1%, with 14% of respondents reporting this fall that they have used the technology for work.

The changes in responses from the summer survey to the fall survey were generally less pronounced than those from the spring survey to the summer, perhaps indicating that the technology has reached a peak in the hype cycle for the legal industry.

The attorneys who have implemented generative AI at work were also asked how they have used the technology in practice and the data show that they are using the technology for more tasks. For example, more respondents in the fall reported using generative AI for legal research (46% summer, 53% fall), summarizing legal narratives (34% summer, 42% fall), reviewing legal documents (20% summer, 34% fall), and due diligence (19% summer, 21% fall).

The data suggest that those using the technology are taking the time to fully explore its usability in the practice of law by testing different use cases and figuring out what the technology is good for—and more importantly, what it isn’t. Granted, how an attorney uses generative AI for work largely depends on how their organization has responded to the technology and how they allow their staff to use it, which respondents were also asked to report.

If you are having trouble viewing this interactive graphic, please enable third-party cookies on your browser. Terminal users, please click here.

In-house legal departments continue to outpace law firms in adding generative AI tools into their tech-stacks. A quarter of in-house attorneys (25%) reported that their organization has purchased or invested in a generative AI tool, and 17% have said their company has developed their own—both of which are more than double what their law firm counterparts reported. Only 12% of firm respondents indicated their firm has purchased generative AI, and only 6% reported that their firm has internally developed the technology.

Uncertainty around organizational response to the technology has increased for both law firms and corporate legal departments, with more respondents—both in-house and law firm—selecting “not sure” in the fall (22% in-house, 28% law firm) compared to the summer (10% in-house, 21% law firm) when asked how their organization has responded to AI. This was an interesting change, especially given the consistent spotlight on the technology.

Apart from uncertainty popping up at an organizational level, these developments make sense from an industry perspective as it relates to technology adoption. Although widespread implementation has remained relatively static, it’s clear from the data that the technology is still being explored and the risks are being considered.

With President Biden’s recent executive order requiring more transparency from AI developers and calling for more protections for consumers, it’s likely we’ll see even more movement in the legal industry surrounding the technology in the coming months.

Bloomberg Law subscribers can find related content on our Surveys, Reports, and Data Analysis page, Legal Operations page, and In Focus: Artificial Intelligence page.

If you’re reading this on the Bloomberg Terminal, please run BLAW OUT <GO> to access the hyperlinked content, or click here to view the web version of this article.