I was slow to adopt any generative AI tool usage, not because of the usual reasons (AI hallucinations and word salad), but because I didn’t need it. Years of experience, a fast typing speed, and a client service mentality had gotten me where I am today, so I saw no reason to pivot to a tool that had serious application questions. Now I do.

Tailored software applications using large language models (LLMs), commonly referred to as generative AI, have utility, but how an attorney uses them makes all the difference.

I’ve written blogs about whether or not you want your lawyer using AI, the implications for GCs on routine contracts, and tips for working with counter-parties who rely upon AI. This blog, however, is focused specifically on how a business attorney can optimize the use of AI to provide better client service with increased efficiency.

First, a word on ethics and mistakes – a topic I haven’t written on before because there is a wealth of material out there, including guidance by the American Bar Association. Clearly the tools must be used ethically, but there are numerous ways that attorneys can make mistakes using them, some of which can cross ethical guidelines (especially the duty of competence). With that caveat, here are some of the ways I have found that generative AI can support a business attorney’s practice:

• How can lawyers use AI for an initial document review? With correct prompting, a tailored AI tool (they are not all built for this) can review a document for comments and efficiently spot check issues. While not a substitute for a full read-through of your own (since AI is not actual intelligence and will miss things) it can help an attorney catch (a) the low hanging fruit, and (b) some things an attorney might not have caught. Having tested this, I will say that in my experience it also creates a bunch of word salad, so it’s critical to understand it as only one step in the review process, not a substitute for a complete review. Still, it can be helpful.

Clients are often doing this anyway, frequently not with software tailored to protect the attorney client privilege (another area of concern), and it will help an attorney to understand how the client is generating their questions.

• How can AI help lawyers revise and clean up legal documents? As lawyers re-read documents, cleanup is commonly required (definitional alignment, catching undefined terms, section cross references, redundancy, term inconsistency, general wordsmithing, etc.). Those cleanup steps are time consuming and often require a complete read-through of a document an attorney has reviewed several times to catch those types of issues. Generative AI tools can speed up that process by (again) catching the low-hanging fruit. After all, if they can catch 90% of the issues, the other 10% will be easier to identify. The tools may miss some errors, and more importantly, may flag as errors items that are correct; but with proper analysis, this can significantly speed up the final review process.
• Can lawyers use AI for legal research? There is value in starting with AI, as the LLMs consolidate large bodies of information into a digestible format. However, this requires extra caution; you cannot rely on the direct AI results, but must dig into the links and often beyond, to the associated black letter law. A lawyer who takes it to the next level and is diligent about the process can develop their professional competency in a nuanced area of law by starting with AI tools and developing it from there.
• Can AI help lawyers anticipate a counterparty’s response? I explored this in further detail in earlier blogs where I focused on how to deal with counterparties that use AI, but using AI to anticipate how your counterparty will respond to your email, draft contract, etc., can help you prepare for a conversation. Reiterating my concern: without experience and skill, you may come to rely too much on AI; but sometimes a quick run-down based on quality prompting can make your discussions more effective.
• How can lawyers use AI to make legal language easier for clients to understand? Let’s face it, lawyers can sometimes get lost in the words when trying to convey very technical information to a client. Client service means the legalistic analysis and law being referenced must be made clear and concise for the client’s benefit. AI tools can soften language and, with good prompting, keep the substance intact. This takes effort because you must be careful that AI does not change the meaning or intent of your phrasing. Still, it can be helpful as a second set of eyes.

Optimizing AI usage really means knowing where it supports the attorney and helping clients to recognize that AI (at this stage anyway) would do a very poor job of replacing an attorney. That said, I’ve come to accept the adage that AI won’t replace attorneys, but attorneys who know how to properly utilize AI will be better attorneys than those who do not.

For those who despise AI because of its usage in art, I’m sympathetic to that, but please recognize the law of public domain, and the good AI tools are trained on materials intended for attorneys’ use anyway (like template databanks or analytics that would have been licensed to attorneys regardless of whether they were using an LLM). However, not all tools are created equal. I’m not going to endorse a product in this blog, but I’ve been testing some, and found that each tool does some things better than another tool and some things not as well. It’s important to read not only the application’s intended use, but also its terms and conditions and user agreements, before entering confidential information. Attorneys must take care not to employ tools that can impair the attorney-client privilege or otherwise be used unethically.

Of course, AI is evolving quickly so all of these observations could change as do the times.