How has AI affected your efficiency and productivity?
DB: As a search engine AI improves my efficiency because it posits answers and footnotes them with the source for its response that I can use to verify the response. For other purposes, I can't say that it has improved my efficiency YET because I'm still in the learning curve stage. I spend a good bit of time learning how to use the tools.
ME: It has sped up some of my work, especially when I'm stuck crafting words or outlining writing. But, I always leave time to review my work before submitting it or publishing it.
SG: It allows me to do first drafts and summaries quickly, so it helps with procrastination.
AK: As I am a trial lawyer, I receive many sets of documents, mostly medical records, in PDF format. Before Copilot (BC), I would scan in the documents, OCR them, and then do an advanced search for words and phrases. Since I subscribed to Copilot, now I ask the program to summarize the documents, ask Copilot about certain injuries and dates, and also provide me bullet points regarding specific issues in the records. Copilot also assists me in summarizing Deposition transcripts, searching for inconsistencies in testimony, and even providing pages and lines of testimony I am looking for.
AP: While it does not really save me any time in the overall drafting process when it comes to motions and briefs, it has often contributed to improving the eloquence or clarity of my papers. However, when it comes to simpler drafting, anything from demand letters to case summaries, it has definitely reduced the time spent by at least 50%.
RT: I don’t use these products every day, so my proficiency is still low. However, as I learn new ways to use AI products and get comfortable incorporating them into my workday, I can see my productivity increasing.
Which AI products have you found most useful, and why?
DB: The image generation by Copilot has been good for presentations. The drafting capability for emails and articles is great for overcoming "the blank page." It also takes some of the drudgery out of the process of editing and formatting. It is really good at improving what I have written, although it may require multiple iterations to get the tone and content to my satisfaction.
ME: I have discovered ChatGPT, DALL*E, and AI-Writer to be useful tools. The key word being "tool." These products are all great tools to help with writing, outlining, and even generating neat pictures. But they do not replace the author's responsibility to check one's work.
SG: The summarizers and generic AI are good just to ask questions. It's like souped-up Google. I found myself using it to try and explain certain topics to my undergrad students sometimes when I found myself explaining like a lawyer instead of a professor.
AK: For the price, I’m a Copilot devotee. The reason I find Copilot to be useful is that it works seamlessly with Office 365 and Microsoft Edge. For PDF documents, once you have performed Optical Character Resolution, I right-click the file in my browser and “Open With” Microsoft Edge. The document is now in Edge and Copilot is open. The first thing I do is ask Copilot for a summary of the document and I’m off to the races. If there is something I would like to incorporate into a document such as a demand letter or letter to another party, I can just cut and paste what Copilot provided me into the document.
AP: Claude. See my answer to question 1.
Are there any AI products that you've tried but found less helpful? If so, why?
ME: I'm sure I've tried other AI products and found the respective results unuseful, i.e., not as good of a result as those I mentioned above. When the results are not polished enough to be of any assistance, I move on – but recognize I may come back and try them again after they have had time to improve their product.
SG: The ones that create images are fun but weird, and it's getting so more of the AI tools are pay-to-play, so it's not worth signing up for anymore.
AP: As I mentioned, I've tried various legal research platforms, and have yet to be impressed with any results. I was also less than impressed with ChatGPT when it came to legal work, as it would invent too many facts and produce hallucinated citations too often.
RT: Finding time to try different products is not easy. I’m using TheresAnAIForThat.com to research products.
What challenges or ethical considerations have you encountered while using AI in your legal work?
DB: I want to make sure that I am not revealing confidential client information or inadvertently training AI based on client information that could have detrimental impacts. I think some of the software that helps legal departments analyze and organize their existing contracts could be invaluable. Based on discussions I've had with in-house lawyers, some legal departments in the past did not find that their results were what they had expected. The advances in AI may help users employ the tools more effectively and get better results. I think it will be important for them to understand whether and how their confidential information is being protected, however.
ME: I haven't experienced any, as I am very cautious about how I use AI in my work. I don't input my own or client's PII when I have AI generate for me. I definitely don't have it research topics or draft briefs without checking case citations. I almost always check an opposition's citations – it is amazing how many errors I would sometimes find. Likewise, I would check a clerk's citations. Thus, why would I not check an AI's citations?
SG: It's like working with a non-attorney. You have to check their work. Just like you'd never write a brief without checking to see if it is still good law, you never should take AI at face value.
AP: Well in essence there are two. First is the potential for letting privileged and/or confidential information into the internet. To address that, start by reading the Terms of Service on any AI platform you're going to use before you put anything confidential in there. Some platforms may not be acceptable at all for that sort of thing. Others may be acceptable only with a paid subscription. As with selecting a cloud storage service, a practice management system, or any other key service, make sure the Terms of Service match with your intended uses. This issue can also be addressed by simply not placing any confidential facts of the sort into the generative AI and instead changing the names and basic facts enough to produce an anonymized hypothetical that analogizes the case so it can produce the work that you need, with only minor changes needed on your end afterward. The second is the potential for error. This, again, is nothing new. Much like when you have to supervise your unlicensed JD drafter and sign your name without reviewing the fabricated brief you're handed, it's your license that's on the line. As the attorney, making sure you cross-check, review and edit the work product before deeming it final, prevents any ethical issues.
RT: I don’t completely trust ChatGPT so double check specific facts it spits out. I think that is part of our competency. Some products, such as Visalaw GEN provide citations to their sources. That transparency is very helpful to maintain our professionalism.
If you are not currently using AI in your daily work, have you considered exploring AI tools? If so, which ones? If not, or if you have decided against using AI, what reservations or concerns do you have?
DB: I want to try using AI to assist me with using Excel because my Excel skills are so rudimentary.
ME: I am curious about various legal research platforms' use of AI. It will be interesting to see how Lexis, Westlaw, and Fastcase, will be using AI to help lawyers with their research. I am also exploring more on using AI to help repurpose prior, current, and future posts on my blog and podcast.
RT: I don’t have any reservations about using generative AI in my work. I was practicing law when the Internet took hold and when email was launched. There were ethical concerns that held some lawyers back then too, yet the vast majority of lawyers successfully navigated those new technologies without incident. I believe the same will be true of this new technology.