Legal Tech Trends Newsletter: #2
Pre-mortems, Harvey, Aleph Alpha, cyber attacks, SALI adoption, and much more!
Hello, and welcome to the second edition of the Legal Tech Trends newsletter!👋
A very special shout-out to the 111 new joiners since our last edition. You can expect a fortnightly shot of the top Legal Tech trends, insights, and my personal recommendations.
This newsletter is only possible because of the great work shared by others in the industry. I built LegalTechTrends.com to help surface that content. Here I share what resonates most with me.
My point of view is shaped by many sources and experiences, including my conversations with legal industry professionals, so please do feel free to provide feedback on any topic I share.
I hope you get value and take away at least one insight from reading each edition. 😊
✍️Titans Insights
I’m a fan of the great work being done by Nikki Shaver and Jeroen Plink through their extensive Legaltech Hub directory. I write a monthly column for the website. This month’s article focused on pre-mortems, a project management technique involving imagining failure to save you from a project disaster. The article is only available to premium users on the website, but you can also access on the Titans website: LINK
🚀OpenAI
Once again, OpenAI stole the headlines over the last two weeks.
Allen & Overy partnered with Harvey, a start-up OpenAI invested in, and have rolled out GPT access to more than 3,500 lawyers across 43 offices. The announcement was widely covered, including Wired, LawNext and Legal IT Insider.
My top three take-aways from the coverage:
🏃♂️Pace of delivery: A&O first heard about Harvey in September and rolled out the product to 3,500 lawyers in mid-Feb. Kudos to them for moving at pace!
🔐Data: They are not using client data. The Infosec team will not allow the use of client data until they are “convinced that the data can be ring-fenced and protected from other use”.
🎯Accuracy: The models have been fine-tuned for A&O, but they still experience hallucinations. “You must validate everything coming out of the system. You have to check everything.”
This is exciting news, and it’s great to see law firms embracing new technology, but let’s balance our optimism with sufficient scepticism and suitable rigour.
I’m exploring a number of use cases for this technology, so I care deeply about overcoming these challenges early.
🤔 Three topics I’m pondering:
Data Usage
Where are the models hosted? How is the data being used? How can we comply with infosec and data privacy standards?
Accuracy
How can we detect hallucinations/inaccuracies/bugs? How can we best reduce these inaccuracies? What are the risks of surfacing inaccurate information?
Explainability
How can we explain the output generated? Where did it come from and what has informed it’s inclusion? Are we comfortable explaining this to clients?
Software engineers are seeing amazing results using OpenAI to assist them write code, but the origin of the code matters less for them.
The importance of this current challenge was highlighted to me through discussions with Jack Shepherd and Alex Herrity. I also invite others to provide feedback on this topic.
📢 I’m currently ‘crowd-sourcing concerns’, to surface key challenges to tackle early and identify the most suitable cases we can use the technology on right now. Legal is a broad domain, so it’s important to source many perspectives! Feel free to respond directly to this email or message me on LinkedIn.
🗞️ Other OpenAI news
When you use OpenAI models via their API they may use the data provided to train their models. There may be ways to opt out of this. But be warned. You should be very clear on how OpenAI uses the data before using it for anything sensitive. LINK
OpenAI isn’t the only large language model in town. Aleph Alpha claim their largest model is trained on more parameters than OpenAI ( 300bn vs 175 bn); they can deploy in any environment (including on-premise) and provide sources and citations for where an answer has come from. LINK
Linklaters suggest seven rules of the road before integrating ChaptGPT into your business LINK
✅ Standardisation
Thomson Reuters is expanding its support of the Standards Advancement for the Legal Industry (SALI) Alliance's universal taxonomy by adopting open standards for Accounts payable/billing and firms in Legal Tracker Advanced, Matters in HighQ via iSheets, and Contracts in Document Intelligence. This is another promising step to improve interoperability and consistency in the legal industry. LINK
⏰ The Billable Hour
Traditional legal pricing models can pose major challenges to introducing new technology. Data is scarce, but most work is still billed by the hour. What if we capped billable hours? What if we linked price to service levels? LINK
💥 Cyber Attacks
A cyber attack hit Troutman Pepper. LINK
The New York City Bar suffered an attack last December. It appears they may have tried to keep it under wraps. LINK
🎙️Podcasts
Lawnext: Ep 194: Paralegal Kathryn Tewson On Her Quest for Accountability from DoNotPay
DoNotPay, a self-help legal site, is pitched as the world’s first robot attorney but investigations suggest it’s actually document automation and human researchers. Bob Ambrogi discusses the saga with Kathryn Tewson. LINK
Johannes Scholtes: AI Is Finally Here. Now the Hard Work Begins for the Legal Industry (TGIR Ep. 191)
The promise of AI has been around for decades, but it is the last three months that has finally caused an awakening so forceful, that even the legal industry understands it needs to be ready for the upcoming Age of AI. LINK
📑 Reports
The Thomson Reuters Law Firm Financial Index (LFFI) dropped to an all-time low in Q4 2022 due to declining client demand and tall baselines from the busy Q4 2021. This caused productivity to be at an all-time low and expenses to increase. LINK
Adjacent Industries
📈 Tech Trends
Renowned industry analyst Benedict Evans has released his annual presentation exploring macro and strategic trends in the tech industry. Much of the presentation focuses on media, advertising, and e-commerce, but it also covers generative AI. Those in innovation teams may like this read. LINK
Aaand that’s a wrap for the second edition. The next one will drop on Friday, 10th March. In the meantime, to satisfy your curious legal tech mind, explore LegalTechTrends.com.
Thanks for reading. All feedback and suggestions are most welcome.
Have a great Friday!