[EDRM Editor’s Note: This article was first published August 4, 2023 and EDRM is grateful to John Tredennick and Merlin Search for permission to republish. The opinions and positions are those of John Tredennick.]
Three of my favorite authors, Ryan McClead, Greg Lambert and Toby Brown of Three Geeks and a Law Blog, wrote a fascinating three-part analysis on the impact of GenAI on law firm profitability that should be required reading for everyone in the legal profession.
The series started with an in depth look at the oft-cited Goldman Sachs article suggesting that the legal profession would shrink by as much as 44% due to the productivity enhancements (or job stealing depending on your views) of GPT and its Large Language Model progeny. 44% of Investment Bankers Think They Can Make Lots of Money Off of Attorney Insecurity (AI).
Their second post took a look at different data and projected that the impact would be more like 23% (still a significant impact but half of the Goldman Sachs gloom and doom numbers). Generative AI Could Reduce Law Firm Revenue by 23.5%.
Their final post then offered insights and advice on how to ride this new wave of AI technology and end up standing on your board. AI-Pocalypse: The Shocking Impact on Law Firm Profitability.
This was serious work. At 5,600 words and 32,000 characters, there was a lot to read and digest. But the topic is important to the legal profession and the authors’ conclusions are worth serious consideration.
But I am too busy billing hours to read something this long.
Right, me too. So I took the easy way out and simply asked Claude 2 to summarize the article for me. Why Claude 2? Because GPT 4 kept rejecting my prompt because the underlying articles were too long. I even tried submitting the posts one by one. Only the third one made it through. I could have tried further trimming but I ran out of time.
For those newer to GenAI, GPT 4’s context window (the space for analysis) is 32,000 tokens (about 2,800 words). Claude 2, the leading competitor to GPT 4 in our judgment, can handle 100,000 tokens in one go (enough for an entire book).
So, if like me you feel too busy to read all three parts, Claude 2 can help.
Read the entire article and view Claude 2’s take on the three-part blog post here.
