AI for eDiscovery in Small Cases

AI for eDiscovery in Small Cases – Do We Need It?: eDiscovery Best Practices

Tom O’Connor gives his answer to whether we really need AI for eDiscovery in small cases in the first sentence in his article. But he says a lot more after that.

Tom’s article (Do We Really Need AI for eDiscovery In Small Cases?, written for the New Orleans Bar Association and available here) is a comprehensive look at the issues and considerations related to AI for eDiscovery in small cases, written in a matter-of-fact, straightforward writing style that I really like (you can just picture Tom saying the words in the article in a conversation with you).

Tom starts with the discussion with definitions – of small firm, AI (including non-legal, legal and litigation uses of AI), generative AI (accompanied by the infographic 120 Mind-Blowing AI Tools that I covered here) and large language models (accompanied by a “WARNING … THIS IS WHERE THINGS GET TECHNICAL” message and a picture of Will Robinson & the robot from Lost in Space).


The LLM discussion is the most in-depth, but it’s also one where Tom acknowledges just how technical the topic is when he says: “The size of an LLM is important because certain capabilities only emerge when models grow beyond certain sizes. Ione study, by Jerry We and Denny Zhou, of Google Research, showed that some mathematical and word skills appear in LLMs only when they grow past certain sizes and have been trained for long enough (measured in training FLOPs).”

And then follows that up with: “I won’t even go into training FLOPs. It would only confuse me while watching the Pelicans play the Warriors.” {Now you understand the reason for that image at the top!}

Tom also dives into several “other” issues to consider when discussing generative AI including the proprietary nature of some databases, hallucinations, privacy, and cost. Pretty important issues! Tom covers them to give the reader a sense of awareness to be able to address those issues.

One of the great things about this article (and Tom’s writing in general) is that he includes links to numerous resources and quotes from them as well. So, a lot of the discussion isn’t just his opinion on these matters, it’s what he has gathered from various reliable sources in our industry.


He even references my post The Blackberry Lesson for eDiscovery Providers, which is how I found his article when I did some testing yesterday of GPT-4’s “Browse with Bing” feature by asking it to synopsize my post and it hyperlinked to his coverage of it. That’s why I’m just now covering it, even though Tom wrote it in August, as I just found it yesterday.

Tom also references a 2019 post he wrote with the same advice he offered back then:

“Technology is a tool and the responsibility for using the proper tool in the proper manner resides ultimately with the attorney. As I have said before, we need to keep the attorney in AI.”

“It’s not enough to be aware of AI, we have to understand AI. Or, as the great technologist Elvis Aaron Presley once said, ‘A little less conversation, a little more action please.’”

Well, thank you very much, Tom! See what I did there? 😉

So, what do you think? Do you think we really need AI for eDiscovery in small cases? Please share any comments you might have or if you’d like to know more about a particular topic.

Disclaimer: The views represented herein are exclusively the views of the author, and do not necessarily represent the views held by my employer, my partners or my clients. eDiscovery Today is made available solely for educational purposes to provide general information about general eDiscovery principles and not to provide specific legal advice applicable to any particular circumstance. eDiscovery Today should not be used as a substitute for competent legal advice from a lawyer you have retained and who has agreed to represent you.

Leave a Reply