It goes without saying that one of the most critical functions of a law firm is to train its associates adequately. But time constraints and a lack of consistency, as I have previously discussed, make good, sound training of associates problematic in many firms. However, large language models and GenAI, even open models, may offer potential solutions. Provided, of course, that the firm and its partners understand the risks and benefits of these models and how to use them.

The Shortcomings of Traditional Training Methods

Part of the problem with training young lawyers is that, by necessity, it is very experience driven. To paraphrase Abdi Shayesteh, co-founder of the training firm AltaClaro, you can’t learn to swim by reading a book. You have to get in the water. Want to know how to take a good deposition? Watch a good lawyer take some depositions and then take a bunch yourself. Want to learn how to pick a jury? Second chair some trials and then try some cases on your own. This learning through experience was how many of us become good lawyers.

This kind of on-the-job training no longer works so well

But this kind of on-the-job training no longer works so well. Clients aren’t willing to pay for associates to sit in on depositions or trials. Clients aren’t willing to let inexperienced lawyers work on cases. Firms aren’t willing to write off associate time. Partners aren’t as willing to miss out on billable time to shadow and mentor associates on how to practice law. And, of course, there are fewer trials.

The result is more dependence on structured training and the use of handouts, videos, and reading assignments. But a three day workshop or a video doesn’t take the place of hands, in the trenches experience.

Enter Large Language Models

So what’s the answer? I recently sat through a workshop on the use of LLMs in litigation. The leader of the workshop, an experienced Miami trial lawyer named Frank Ramos, used open AI systems to show how he could prepare for his client’s corporate designee deposition. The deposition was to be taken by a plaintiff’s lawyer who, in the past, had employed the reptile theory approach to depositions.

After carefully crafting a prompt that contained no confidential or other identifying information, Ramos was able to get the LLM to quickly spit out a series of good questions. In response to another prompt by Ramos, the LLM turned these questions into leading questions. It even supplied some potential answers for the witness.

In another demonstration at the same workshop, Ramos described his case in general terms in a prompt and then asked the LLM for some suggested voir dire questions. The LLM provided a whole series of questions that were most of what I would have come up with on my own. It just would have taken me longer.

The lawyer then asked the LLM what answers he should look for from potential jurors that might suggest a predisposition for or against his case. Once again, within seconds, the LLM provided a series of responses that were all pretty much right on point.

Once again, all of this work product came not from an expensive closed system that many vendors are touting as the only way lawyers can use GenAI. Instead, the responses came from an open, $20 per month service that is publicly available. And none of the prompts, again, contained any privileged or confidential information.

These models, combined with a help of a more senior lawyers could provide training with much less risk and cost than traditional methods

Using LLMs To Train

What does this have to do with training? As I watched this and other previous demonstrations, it occurred to me that LLMs could provide a way to get the kind of hands-on training and experience that associates need. These models, combined with a help of a more senior lawyers, like Ramos, could provide training with much less risk and cost than the traditional methods might generate. A senior lawyer could work with an associate to put together the prompts needed for an inquiry. The senior could show the associate how to structure a prompt without revealing confidential materials. 

The associate could get the results and then consult with the senior lawyer about what might be worthwhile and usable and what was not. Why should some questions be used and others not, separating the wheat from the proverbial chafe. At this point in the process, the two lawyers working together could add into the response whatever confidential information was needed to make the LLM output useful in real life.

The associate would learn how to leverage LLMs to get good results and too brainstorm. Says Ramos, “LLMs provide firms a new resource to train their teams to explore ideas, imagination and approaches by using them as brainstorming tools and sounding boards”.  

At the same time, young lawyers would learn how and why some questions and answers provided by an LLM are good and some are not. In essence, substantively learning “by experience” in a safe and secure way. Because the process is efficient and can be replicated or scaled over many matters, the associates gain confidence and experience. This kind of process would also enable firms to employ institutional knowledge and consistency across the associate ranks.

Finally, as Ramos notes, “Young lawyers often are intimidated to go to the partner’s office with yet another question.  Now they have LLMs to explore those questions and have someone, albeit a collection of algorithms, to help them think through issues in their case”. 

The Payoff

Of course, this presupposes that senior lawyers take the time to master the use of LLMs themselves. But that’s something they should do anyway to represent their client adequately in the long run. And it need not be that hard. Effective use of LLMs takes some communication skills and practice. Yes, it would require some investment by the senior lawyers. It would require a firm commitment to a training process. 

But for the firms that institute this kind of training in a studied and systematic way, LLMs hold lots of training promise. But the long-range payoff could be significant. Better overall associates and perhaps a willingness of clients to allow younger lawyers to do more when they see their value and experience level. And even pay for their time. Happier and more confident associates.

It beats hiding your head in the sand when it comes to training…and LLMs..