skip to content

Judge Iain Johnston on the New Era of Ediscovery and His Hope for Generative AI

by Justin Smith

Reading Time — 12 minutes

Judge Iain Johnston has seen practically every corner of the legal system.

After starting out in the Illinois Attorney General’s office, he moved to private practice, stopping at Am Law 25 firm Holland & Knight, among others, before setting up his own shop.

Now a judge in the United States District Court for the Northern District of Illinois, he’s made a name for himself by overseeing cases that involve complex ediscovery issues, using the experience gleaned throughout his career to help create informed decisions.

As generative AI continues to be utilized by attorneys and litigation support professionals, those complex ediscovery issues are likely to become even more involved. Whether it’s during document review or in the courtroom, generative AI is already being used in ways never thought possible just a couple years ago.

Judge Johnston spoke with Everlaw about generative AI’s exciting potential, what law schools need to better prepare their students, and why he thinks litigators are still scared of ESI.

How has the discovery process evolved throughout your career? Has your perspective on discovery changed since you've moved from the bar to the bench?

I became an attorney in 1990, so I have watched the discovery process evolve considerably.

When I was at the Illinois Attorney General's Office, I had no paralegal support. I did all my own discovery. I wrote up all the document requests, interrogatories, and requests to admit.

Discovery was produced and reviewed on paper. It was a tedious and time consuming process, but spending time personally rolling around in the documents was also a tremendous learning experience. I didn't need to rely on somebody else telling me what was contained in the documents. I personally identified the hot docs—both good and bad, and I wrote up the privilege logs, too.

Despite the massive amount of time invested in this process, it gave me control and mastery of the case. Pleadings and depositions were better because I was able to address discrepancies between the recollection of a client or witness and documents before allegations or answers were reduced to writing or witnesses testified under oath in depositions.

Obviously, I wasn't billing a client for this time, so I had the freedom to spend as much time on this work as I wanted. I spent a lot of weekends in the James R. Thompson Center in Chicago.

When I transitioned to private practice, some of that changed. I still liked to review the hard copy documents because that was the best way for me to learn about and understand the case. Then technology ramped up both in the production of discovery materials as well as the review of discovery. The volume of documents to produce and review exploded, and the technology side of things was slow to catch up.

I remember when the ability to OCR documents came along and being giddy that I could conduct word searches for attorneys' names to help with identifying privileged documents. Then off-shoring discovery began. I recall a local government client grumbling about off-shoring discovery review until my partner and I explained how much it would cost for an army of associates to review the documents.

Technology continued to advance with the advent of TAR. When I first learned of TAR, I was very skeptical. My partner went to a CLE course about it and came back to the office, telling me how great it was. Like a lot of attorneys, I was still skeptical, but my views changed when I saw the studies validating TAR. Now TAR is accepted and expected in the appropriate cases.

I'm very curious to see how generative AI (however it’s defined) will be used in the discovery process, but I'm even more interested in seeing how the use of generative Al tools that are still in development are going to be validated. Obviously, as technology advances, including creating endless communication programs, the discovery process will need to evolve.

Judge Joy Conti previously identified more than 30 communications programs, and back in 2019, Judge Katharine Parker noted that the proliferation of electronic communication can result in collection and/or review of 100 gigabytes of data.

Think about all the new communication programs that have been developed and are being used since 2019. I worry that we are quickly reaching the point at which meeting the goals stated in Rule 1 and the requirements of Rule 26(b) will be unsustainable.

As to changes in my perspective regarding discovery since transitioning to the bench, I will freely admit that I am a stronger proponent of cooperation among counsel. Don't get me wrong, as an attorney I had excellent working relationships with most—but certainly not all—of my opposing counsel. In fact, it wasn't unusual to have lunch or grab a beer with certain opposing counsel, and I often referred cases to those attorneys and they referred cases to me.

You’ve also served as Chair of the Seventh Circuit Electronic Discovery Pilot Program Committee. Can you speak a little more about your work with that, and the sorts of results you’re seeing?

I was the Chair of the Seventh Circuit Electronic Discovery Pilot Program, but I was the last chair and oversaw its wind down. The program wrapped up over a year ago.

The program was outstanding and the people involved should be proud of what they accomplished. It produced many useful tutorials and documents to help attorneys and courts during the early days of ediscovery, and the free continuing legal education programs were top shelf.

It was successful in large part due to the support provided by the Northern District of Illinois, particularly the support of then Chief Judge James Holderman and then Magistrate Judge Nan Nolan, as well as Judge Mary Rowland and Magistrate Judge Sheila Finnegan.

Members of the program included many of the pioneers in the ediscovery field, such as Maura Grossman, Chris King, Kate Kelly, Martin Tully, Magistrate Judge Beth Jantz, George Bellas, and many more.

In another case, City of Rockford v. Mallinckrodt ARD Inc., you adopted the parties’ proposed order establishing the production protocol for ESI with the inclusion of the plaintiffs’ proposal that a random sample of the null set will occur after the production and that any responsive documents found because of that process will be produced. You also mentioned in one of the footnotes that litigators shouldn’t be “scared” of ESI. Now that we’re a few years removed from that case, do you still think some litigators are? Why so?

I do believe that some—if not many—litigators are still scared of ESI, and I wish I knew why.

Maybe back in the day it might be said that litigators were scared of ESI because it was new, but it isn’t new anymore, and hasn't been for a long time.

In fact, I'm baffled by the fear and resistance relating to ESI. The legal profession is constantly evolving in nearly every way. Legislatures pass new statutes at alarming rates. The rules of civil procedure are amended constantly. New understandings of human behavior are uncovered all the time. And attorneys routinely adjust to these events.

I sincerely hope that it doesn’t take regulators charging attorneys with violating their ethical obligations to stay current on technology, malpractice cases being successfully litigated, and attorneys and clients feeling the sting of sanctions for counsel to care about and understand ESI.

In your opinion, why don’t regulators charge attorneys with violating their ethical obligations to stay current on technology? What’s something (if anything) the courts or judges can do to help this issue?

One, it's relatively new. The ethical role of being competent with technology is relatively new in the scheme of things. I think that probably has an effect on why you don't see it.

"There are many core competencies related to ediscovery that are not given priority, but the first step in addressing this failure is for the legal profession to care about ediscovery and know that it should care about ediscovery."

Second, and this is just intuition, I have no evidence to support this, but I don't think the attorneys at the regulatory commissions have a lot of experience with ESI either, and they might be uncomfortable charging somebody with something they don't quite know also. They might feel like they're over their skis.

And again, I don't have any evidence that supports this, but I have a feeling that this might be part of the decision-making process. I have never seen discipline brought for violating that provision of the rules of professional conduct.

Have changes to spoliation rules impacted any disputes brought before you? Do you think the rule is working to ensure more certainty and proportionality around spoliation sanctions? I’m thinking of this both generally and in the context of the memorandum opinion you issued in Marvel Snider v. Danfoss LLC Inc.

Not really.

The 2015 amendments to Rule 37(e) have given with one hand and taken away with the other. Rule 37(e) has provided certainty in some respects. For example, we now know that for sanctions (as opposed to curative measures) to be imposed, there must be a finding of intent to deprive the other party of the use of the evidence. This was the law in the Seventh Circuit before 2015, but was not the law elsewhere. Litigants and parties now have uniformity with this standard.

But the amendments raised a host of other issues, including which party bears the burden of proof as to the various aspects of Rule 37(e), and what the quantum of that burden is. The amendments have also highlighted a division among judges as to who should determine the elements of prejudice and intent: the judges themselves or the juries hearing the case. To the extent that the amendments were intended to decrease the number of spoliation motions, that goal doesn't appear to have been realized. I have a never-ending stream of spoliation motions.

Unfortunately, attorneys—and even some judges—haven't caught on to the amendments to Rule 37(e). In fact, during a recent trial, an attorney raised the issue of alleged spoliation of ESI. So, I asked the attorney if he was making a Rule 37(e) motion in the middle of trial, which is problematic for many reasons. The attorney responded by stating that he was unfamiliar with Rule 37(e). And take a look at many of the opinions being issued by courts. Many still cite pre-2015 case law, including case law explicitly rejected in the Advisory Committee Notes. Some recent ESI spoliation decisions don't even mention Rule 37(e).

There are many core competencies related to ediscovery that are not given priority, but the first step in addressing this failure is for the legal profession to care about ediscovery and know that it should care about ediscovery.

The problem starts in the law schools and then continues into practice. Few law schools teach ediscovery, and from what my clerks tell me it's rarely, if ever, mentioned during their legal education. This is shocking. Law schools that don't teach ediscovery are doing a disservice to not only their students but to the future clients of their graduates.

Each year, I review hundreds of transcripts from prospective clerk candidates. I believe I have seen less than five that have taken an ediscovery class or even a class that touched on ediscovery.

I read a recent article by Craig Ball. Because it was written by Craig Ball, it was both humorous and sad. In the article, he described his attempt to explain to a law school administrator the importance of teaching ediscovery to the students. Unfortunately, it sounds like his message fell on deaf ears.

But there are exceptions. Jay Yelton teaches an ediscovery class at Michigan State University. I've seen his syllabus and it's first rate. I've spoken to his students each semester and it is clear that they really want to learn about ediscovery. Keith Chval teaches ediscovery at the University of Illinois-Chicago. Back in the day, when I was teaching pretrial practice, I'd have Keith talk about ediscovery and tell my students to take his class. Maura Grossman also teaches ESI, which should be a must-take course at any institution that offers the class.

And some firms clearly care about teaching their new associates the importance of ediscovery. One of my former clerks went to Jones Day after his clerkship, and within the first month or so, he was provided with ediscovery training, including a lesson on the ethical ramifications of not being proficient in ediscovery. I was very glad to see that he received this instruction.

Why don’t you think schools see the teaching of ediscovery as a priority? Do schools feel the onus should fall on the law firm that hires them?

I wish I knew the answer to that. Ediscovery has been around for two-plus decades, so I don't know why law schools haven’t fully engaged with it.

Maybe they think they'll leave it to the firms to provide that education. I think law schools do that with some things, but it really is a disservice because think about how many law school graduates hang up their own shingles. They don't go to a firm. They don't go to Jones Day and get this great training. So if that's the belief, then the premise is wrong.

"I think most district courts need to address the onset of generative AI with a local rule, and while some have tried, most haven't addressed it."

There's an old saying with cyclists, that there's a group that loves cycling because of the mechanics and the physics, and there's the group that loves it despite all that. They don't care about the gear ratio, but the engineering geeks love all that stuff.

I don't know if there's that same disconnect in the law, between the technology people and the academics, but it seems like there might be.

I have at least some sense that in the last couple of years, since generative AI has now come to the forefront of not just law, but also society, that there’s a prevailing idea of why should we teach law students about ediscovery when generative AI is going to take it over anyways? What's the point? I think that's out there to some extent. It doesn't seem like a good reason not to address it because they aren’t mutually exclusive in my mind.

Maybe they think it's like doing research with books when everything's online. But it's amazing how few law clerk applicants I get that know ESI or are even aware of it.

You issued a sort of light-touch standing order late last year regarding the use of generative AI, basically suggesting cautious use of it, and reinforcing the duty to competency that already exists. Why did you feel the need to issue this order?

I issued the order for three reasons. First, counsel was questioning me about my views on generative Al in filings. Second, I wanted to contribute to the discussion. I was a bit concerned that some of the more onerous standing orders were becoming the norm. In my view, those onerous orders were unworkable and not fully thought through. Of course, that's just my view, which is often incorrect.

I believe I was one of the first to issue an order with this view. Since then, Judge Martin B. McGee has entered a similar order. So now at least I'm not alone.

Third, it never hurts to remind counsel of their duties.

You mentioned part of the reason you issued this standing order was to contribute to the discussion. What do you think judges should be contributing?

We're so insular as judges. We're kind of siloed.

I think most district courts need to address the onset of generative AI with a local rule, and while some have tried, most haven't addressed it.

I've used the term balkanization before, when referring to this, where one judge does this, and one judge does that, and everything becomes so fragmented and disjointed.

Then I saw the trend of creating these unworkable and restrictive orders. And I thought, maybe a different voice might help the discussion process, because we don't really discuss these collectively across the federal judiciary. There are, hopefully, great uses for generative AI, and if there are such restrictions placed on it, the use might get unnecessarily stymied. That consequence will be detrimental to some of the people who could really use it.

As far as counsel using generative Al at trials, it is and will be an exciting time.

Generative Al is a powerful tool, but like all tools it can be misused. An Allen wrench is a wonderful tool, but it shouldn't be used to drive a four-penny nail. It might work eventually—but there will likely be a lot of collateral damage. And this metaphor assumes a legitimate and proper intent. Misusing a tool with less-than-ethical intent only creates more problems.

"Judges and attorneys should be curious about generative Al's applications and certainly should not be afraid of it, but, as I mentioned, they should also recognize that generative Al is a tool that can be misused."

I do anticipate generative Al to be used much more often to create things like demonstrative evidence or what may soon fall into the category of “illustrative aids.” Think about how generative Al could be used to create a presentation in an excessive force case, for example. But with the increased use of generative Al-created demonstrative evidence, I anticipate an equal increase in the challenges to those exhibits. Just as every opinion witness' testimony is challenged under Daubert (even though that shouldn't happen), every demonstrative exhibit will also be challenged (even though that shouldn't happen).

As far as judges using generative Al in chambers, I know several across the country who are experimenting with it for various tasks, such as to create summaries of depositions or to create chronologies. Generally, the feedback is that for these purposes, generative Al is very useful and helpful.

Personally, I have not used generative Al for those types of things. Heck, I turn off "text predictions" in Microsoft Word. But I have used generative Al detectors when I've questioned cover letters submitted by clerk applicants. Practice tip: Don't have a machine tell me why I should hire you.

I would be very interested in testing a privilege log created by generative Al. Think about how difficult it can be to determine whether a document created by an in-house counsel is covered by the attorney-client privilege or work-product doctrine when the issue is whether the advice is legal advice or business advice.

Of course, whatever is produced by generative Al would need to be validated. After all, how is a claim of privilege ultimately decided? A human reads the document, considers its contents in the context and facts of the case, applies the applicable law, and then makes a decision. Would using generative Al in chambers so that I wouldn't have to conduct in-camera document reviews on weekends bring me joy? Sure. Am I planning on doing that anytime soon? No. But if somebody has a validated study on this, please send it my way. I'd like to read it.

I’d also like to conduct a John Henry-type of experiment in my next bench trial. I’d review and summarize the testimony and evidence the old school way, then try generative AI, and compare the results.

As far as generative Al speeding up various processes of the legal system, I'm not so sure. No attorney has ever told me that they needed less time to conduct discovery because they are using generative Al tools. And I think it is unlikely to happen. I don't recall attorneys saying that they could speed up discovery because of the advent of remote, video recorded depositions either. Likewise, I don't recall attorneys saying they needed shorter briefing schedules when they first started using Lexis and Westlaw.

Any reader of this interview should take these views with a department-of-transportation-shed of salt but, for what they're worth, these are my views.

Initially, judges and attorneys should always keep in mind the goals and requirements of the ethical rules and the rules of civil procedure, including Rule 1, Rule 11, and Rule 26(g). What's more, judges and attorneys should learn about generative Al and technology as they are duty bound to do.

Judge Xavier Rodriguez just published an excellent article on Al. Judge Paul Grimm and Dr. Maura Grossman have written on Al, too. I highly recommend reading these and other publications.

Judges and attorneys should be curious about generative Al's applications and certainly should not be afraid of it, but, as I mentioned, they should also recognize that generative Al is a tool that can be misused. Additionally, judges and attorneys should be cautious about various claims regarding generative Al without those claims being properly validated. After all, this is science. And science requires proper validation.

I envision lots of people with hopes of making lots of money from generative Al making lots of untested representations about their generative Al products. Judges and attorneys must demand that any representations be validated.

Justin Smith Headshot

Justin Smith is a Content Marketing Manager at Everlaw. He’s written on everything from financial markets to the product warranty industry for several different companies.