Gaining the benefits of legal AI for in-house counsel

Separate AI hype and reality

Any rational consideration of AI must separate hype and reality. Many headlines promise legal robots but others make clear that the promise of AI will take time to realize. A study by an MIT economist and UNC law professor quantified the potential impact of AI on the legal market. Their paper, Can Robots Be Lawyers? Computers, Lawyers, and the Practice of Law, found that AI would have a limited impact on lawyer employment. 

The main use cases for legal AI today are in eDiscovery document review, contract analytics, interactive legal advice, and legal research. Multiple law firms have publicly discussed evaluating, piloting, or using AI. If in-house counsel are doing so, they have not gone public about it to my knowledge. In-house lawyers should be aware of two other potential AI use cases: new accounting rules that change revenue recognition and move leases onto balance sheets. 

Decision-makers need to consider the use of AI in a broader context of both legal tech generally and the problems they are trying to solve. Since the advent of PCs and local area networks, uptake of core, practice, and management technology by lawyers has generally been relatively slow. A gimlet eye is required to consider whether AI uptake will be any faster or have more impact than the uptake of many other technologies (e.g., document authoring, deal management, or decision trees) or process frameworks such as knowledge management (KM) and legal project management (LPM).

Experiment with and evaluate AI

As I have said on many occasions, however, I would love to be proved wrong about incremental uptake. I’m just looking at the evidence I see today. I encourage law departments and law firms to experiment with and evaluate AI to understand its potential and be ready to spring into action as needed. AI could well be like the advent of the Internet – not important until fairly suddenly, it is.

For AI evaluation, In-house counsel must consider both the technology itself and the potential savings. Estimating savings is no easy task. Steps include: identify appropriate tasks, measure current unit costs, determine possible future unit cost with AI, run pilots to test, ensure adoption if successful, and track metrics regularly for potential adjustments. Evaluating technology requires empirical testing. Trying to compare back-end algorithms is a task best left to computer scientists and computational linguists. And it’s not clear they will agree. Keep in mind that rigorous empirical testing requires resources.

As the General Counsel or her procurement colleagues consider reducing legal cost and delivering more value, they must look not just at AI. I always advocate defining the problem to solve first, then look for technology. But for those who want to look at technology to start, they should consider other shiny new baubles, especially blockchain and Big Data. These likely hold as much promise as AI.

Reducing legal cost?

If the problem to solve is reducing legal cost, then in my view, the most effective approach is to avoid legal problems by practicing preventive law. (See my writings on this at #DoLessLaw (Twitter) and Do Less Law blog posts.) Technology can certainly help do less law: AI can advise in-house counsel via interactive systems; blockchain can reduce the need for contracts, and Big Data can likely identify systemic issues that give rise to avoidable legal costs.

In sum, if the goal is to reduce legal cost and legal risk and to deliver more value to in-house clients, then AI may be a big help but careful evaluation and planning are required.