Whereas regulation companies wrestle to hold a lid on AI safety threats and in any other case management legal professionals utilizing AI out of flip, Linklaters has a forward-thinking resolution that extra companies ought to embrace. As RollOnFriday notes, Linklaters places its potential robotic legal professionals by way of a UK regulation examination of their very own design, crafting 50 questions from 10 totally different apply areas geared towards a “competent mid-level lawyer specialised in that apply space” and requested the AI to hash it out.
It’s the Child Bar for robots. So, like, a Nano Bar? No matter it’s, it’s working higher than the California Bar.
It’s not like Linklaters is alone on this. Tech professionals throughout Biglaw carry out checks like these on a regular basis. The excellence is that Linklaters is doing this out within the open. And senior legal professionals graded the responses for substance, introducing vital stakeholders to the AI analysis course of that may get missed when instruments are thought of completely by IT groups and agency tech committees. Which isn’t a knock on the conscientious method agency employees and tech-savvy attorneys deliver to those selections however… generally you’ll want to take these points exterior the nerd circle.
The excellent news for companies is that AI is getting higher:
Linklaters famous there was a “important enchancment” within the outcomes when put next with checks it ran in October 2023. The agency mentioned that the AI fashions are beginning to carry out at a stage the place they need to be capable to help in authorized analysis, resembling offering a primary draft or “as a cross-check on an present reply,” and may be helpful “for duties that contain summarising comparatively well-known areas of regulation”.
OpenAI delivered the best-performing AI mannequin on the Linklaters check. That mentioned, the device solely scored 6.4 out of 10 and bet-the-company lawyering doesn’t grade on a curve. The report doesn’t present Grok’s rating, however primarily based on its said method to authorized reasoning, we’re guessing it made OpenAI seem like William Blackstone crawled out of the grave for a spot of tea and a run at some contracts.
However placing AI by way of the wringer in such a public method alerts to legal professionals that the agency each sees potential in AI whereas nonetheless making the clear case to everybody that AI isn’t prepared for prime time.
This method ought to push again towards the “shadow AI economic system” drawback Hill Dickinson just lately addressed. When companies downplay or outright shun AI, unruly legal professionals are going to begin experimenting with AI instruments on their very own. And that’s how you find yourself with confidential shopper information being uploaded to random servers in hostile international locations. Transparency retains everybody on the identical web page in relation to when and the place AI suits into a contemporary authorized apply.
Linklaters makes AI sit regulation exams [RollOnFriday]
Joe Patrice is a senior editor at Above the Regulation and co-host of Pondering Like A Lawyer. Be at liberty to electronic mail any ideas, questions, or feedback. Comply with him on Twitter or Bluesky in case you’re fascinated by regulation, politics, and a wholesome dose of school sports activities information. Joe additionally serves as a Managing Director at RPN Government Search.
