Tony Thai and Ashley Carlisle of HyperDraft, return to The Geek in Review podcast to provide an update on the state of generative AI in the legal industry. It has been 6 months since their last appearance, when the AI Hype Cycle was on the rise. We wanted to get them back on the show to see where we are on that hype cycle at the moment.
While hype around tools like ChatGPT has started to level off, Tony and Ashley note there is still a lot of misinformation and unrealistic expectations about what this technology can currently achieve. Over the past few months, HyperDraft has received an influx of requests from law firms and legal departments for education and consulting on how to practically apply AI like large language models. Many organizations feel pressure from management to "do something" with AI, but lack a clear understanding of the concrete problems they aim to solve. This results in a solution in search of a problem situation.
Tony and Ashley provide several key lessons learned regarding limitations of generative AI. It is not a magic bullet or panacea – you still have to put in the work to standardize processes before automating them. The technology excels at research, data extraction and summarization, but struggles to create final, high-quality legal work product. If the issue being addressed is about standardizing processes or topics, then having the ability to create 50 different ways to answer the issue doesn't create standards, it creates chaos.
Current useful applications center on legal research, brainstorming, administrative tasks – not mission-critical legal analysis. The hype around generative AI could dampen innovation in process automation using robotic process automation and expert systems. Casetext's acquisition by Thomson Reuters illustrates the present-day limitations of large language models trained primarily on case law.
Looking to the near future, Tony and Ashley predict the AI hype cycle will continue to fizzle out as focus shifts to education and literacy around all forms of AI. More legal tech products will likely combine specialized AI tools with large language models. And law firms may finally move towards flat rate billing models in order to meet client expectations around efficiency gains from AI.
Contact Us:
Twitter: @gebauerm, or @glambert
Voicemail: 713-487-7821
Email: geekinreviewpodcast@gmail.com
Music: Jerry David DeCicca
Transcript
The Boston Legal Design Challenge Update - Jeff Marple, Bob Taylor, and Aubrie Souza
Tracy LaLonde and The Joychiever Journey
Consumer Arbitration Made Easier with FairShake's Teel Lidow
Spotting Bias and Politicization of Local News Sources - Loyd Auerbach and Dave Boitano
Helping Startups Get Their Start - Dr. Jacqueline Walsh
Stephen Embry - The Future of the Law Office Won't Need Everyone to be in the Office
Douglas Ferguson on Doing the Work in the Meeting
Richard Hsu on Why Law Firms Will Need Lawyers to Return to the Office
Using Data Analytics to Tell Your Story with RStudio's Sarah Lin
The Boston Legal Design Challenge with Jeff Marple, Robert Taylor, and Gabriel Teninbaum
Olga Mack on Valuing Your Skills, Reputation, and Determination
Phil Flora on Leopard Solution’s New Gender and Ethnic Diversity Tool
Using Data to Really Know Your Clients and Predict Their Needs - David Kamien
Williams Lea CEO, Clare Hart on Letting Lawyers Do What They Do Best… Practice Law
Andrea Markstrom and the i.WILL Forum. Women Empowering Women
The Law Firm Antiracism Alliance - Brenna DeVaney
Law Firm Culture and Marketing, and How to Market Law Firm Culture - Barbara Malin and Jennifer Johnson
#Barpocalypse - Cat Moon, Brian L. Frye, Stefanie Mundhenk
Yes, And… A Return to KM 101 - Eugene Cipparone
Text, Context, and SCOTUS' Textualism in Bostock - Andrew Koppelman and Sara Harris
Create your
podcast in
minutes
It is Free
The emPOWERed Half Hour
NABOR® TALKS
U.S Property Podcast
Aligned Money Show
The Ramsey Show
Planet Money