I have recently begun experimenting with using AI, which stands for “Artificial Intelligence”, to help me in my day-to-day work. I will explain how I use it and how I don’t use it, but first two things:
- I am not a “tech” person. I rely on technology every day in almost all aspects of my work, but I am more of an “I get by” person, not someone who flourishes with tech. If I have a technology problem, i.e., my emails are not being sent, I can’t open attachments, or a myriad of other issues, my frustration levels spike. Thankfully my law firm has a very friendly and experienced outside technology service that is available every single day. I get by, but at times, it is not pretty.
- We’ve certainly all heard about AI in the news and I cannot discuss or even think about AI without a reference to one of my favorite scenes in Stanley Kubrick’s 2001: A Space Odyssey (1968). In the movie, the astronauts on a spaceship traveling to Jupiter have a disagreement with Hal, a supercomputer who converses with the astronauts. The astronauts conspire to unplug or shut down Hal, but Hal has outsmarted them. It is a timeless sci-fi film and the dialogue between Dave and Hal, in which Hal informs Dave that it will not allow the astronauts to disconnect him as “the mission is too important” is worth watching (“I’m sorry Dave, I’m afraid I can’t do that”). A link to the scene is here.
There are also dozens of other movies in which out-of-control supercomputers wreak havoc on the human population. The Terminator (1984), The Matrix (1999), I Am Mother (2019), are a few that come to mind. It always makes for an excellent plot.
How I use AI in my day-to-day work:
I am a licensed attorney in several states and on any given day, I engage with a wide variety of dense Medicaid and Medicare regulations. This entails working within the notoriously complex Social Security Act, but also state statutes, regulations, policy manuals, etc. It can be challenging and confusing, to say the least. I have extensive experience in this field, so a lot of times, at worst, I have some vague knowledge about a particular regulation, but finding it can be another problem. AI can be great to assist in narrowing down where to find that one regulation I know exists that is directly on point to an issue I am dealing with that day. Particularly if it is some obscure regulation that I really need, this can save me from having to look through dozens of old briefs I have written for case names that I have long forgotten. I will delineate further on this below, but it is imperative that you double-check and verify the information provided by AI. For some reason, possibly related to Hal 9000, AI will lie! As a result, AI may be my first step, never my last.
Some attorneys practice in the same courthouse for their entire careers and lucky them. For those of us who are tasked with a multitude of courts in a variety of states and an exponential volume of regulations, statutes and local rules to navigate, having Chat GPT (or other AI) as a preliminary assistant to find, for example, where the state houses its notice requirements when commencing an Estate proceeding, can be very helpful. Narrowing down a substantial list of potential regulations to a few (dozen sometimes) can save both me and our clients significant time and money in preliminary research where less-frequently visited courts are involved.
The search function is supremely easy to use. There are no Boolean connector tools to memorize (I have always struggled with these). Normal prose is perfectly fine to use or even phrases connected to the topic you are seeking information about. Chat GPT, which I use, is remarkably good at understanding the topic you are seeking information about. Also, it is capable of understanding follow-up requests, for example, requesting that Chat GPT identify its sources for information it provided to you.
What I use:
Chat GPT. Its free! There is a paid/subscription version of Chat GPT, but for my limited purposes, the free version works great for me.
There are also other AI products available to the public. Chat GPT informs me that the following are available as “Chatbots & Language Models”:
- ChatGPT (OpenAI): A conversational AI model available in free and paid versions (GPT-3.5 and GPT-4).
- Bard (Google): Google’s AI chatbot that integrates with Google services.
- Claude (Anthropic): An AI assistant focused on safety and helpfulness.
- Microsoft Copilot: Integrated into Microsoft Office products (Word, Excel, etc.), it assists with tasks using AI.
- Character.AI: An AI platform for creating and interacting with custom AI characters.
What Not to Do:
This is specifically for lawyers, and I can’t stress this enough: never, EVER, use Chat GPT to find case law. When we all started hearing about AI in the news, there was a story about a lawyer citing case law that he found using AI. The “opinions” never existed, they were completely fabricated by Chat GPT. The Judge then asked the lawyer to file copies of the opinions with the court, and the lawyer in question (poor soul) went back to Chat GPT and asked it to write opinions for the cases, and then filed those with the Court. A link to the story is found here and here.
AI likes to “hallucinate” (yes this is the technical term for it), in which it really just makes things up. AI quite enjoys making up case law. I know, I’ve tried it purely for experimental purposes as our firm has a solid AI policy against such types of uses. But yes, the cases it came up with did not appear to exist at all. So for case law, save yourself from having to explain to a Judge that you filed a brief with citations to made-up case law and stick to WestLaw or LexisNexis.
Double check all AI generated information, and never use it as your final work product. First, it’s inherently unreliable. Second, AI is still new, which means AI use cases, ethics, and the legal uses of AI are all still also new. There is a constantly evolving landscape surrounding best use and reasonable use and required citation and disclosure regarding A1 use. Until and unless these outstanding issues are resolved, it really is best practice to utilize A1 much in the same way you might utilize a Google search, but not in the way you’d rely on a personal assistant to compile a first draft of a paper. Last, but certainly not least, your firm, like mine, may have its own A1 policy, which, of course, requires your compliance. If you have questions about using A1 in your professional life, or even personally as a volunteer on a Board or committee, it’s always best to ask the person most responsible for compliance for their guidance.
We’ve all heard about potential replacement of workers by AI and certainly you can imagine that this is a possibility. But for now, I am not at all threatened by Chat GPT or other AI, and I even find it useful to help me to work more efficiently.
