AI Jurors Test Your Case Before the Real Trial

Collin McKee

Welcome back to the Endeavors AI Podcast. I'm Collin McKee, founder of Endeavors AI, where we talk about the intersection of artificial intelligence, law, business, and the new economy that's being built all around it. Today's guest spent more than 30 years as a trial lawyer before an illness took him out of the courtroom — but not out of the fight. He built a focus-group and mock-trial consultancy from the ground up, and when COVID shut down the physical venues he'd relied on for years, he partnered with AI developers to build something new: AI-powered synthetic jury research that defense firms, plaintiff firms, and major insurance carriers now use to test their cases before ever facing a real jury. Jerry Goldstein, welcome to the show.

Jerry Goldstein

Good morning, thank you. Glad to be here.

Collin McKee

So Jerry, let's start at the beginning. Thirty-plus years trying cases, and then a health issue takes you out of the courtroom. Walk me through what that transition actually looked like for you.

Jerry Goldstein

Well, I was significantly unable to continue my trial work for health reasons. But I was so involved in case management and case preparation that attorneys would be calling me and asking for my opinion and my thoughts about different cases — friends of mine and people in the trial bar. When I started out, the trial bar was much smaller than it is today, and for about 10 years I was specializing in medical malpractice cases. You get to know the groups of people you're working with — it was the same attorneys all the time, the same firms. While we were collegial colleagues at the bar, we were nevertheless constantly battling each other. So when I stepped back from the active trial bar, I began to get inquiries about strategies and preparation.

I had had a couple of experiences doing live focus groups, and I thought they were an absolutely marvelous tool to prepare a case — especially if it had significant value and significant factual issues. So I started a company in New Jersey where we began to do live focus groups, and we built a small courtroom within our facility. We had attorneys from every walk of life and every kind of case. And then I received a call from one of the major hospital centers, who had heard about what I was doing and said I had been successful against them many times — would I consider doing work for them? And I said absolutely. And I loved it. Some of the defense firms then became friends of mine, and that's how the whole thing really expanded.

We were doing probably a hundred or so live focus groups a year. I had my own unique way of doing them, in which I presented both sides of the case to the panel in a summary fashion. I did the recruitment, I did the hiring, and at that time it was relatively inexpensive — for seven or eight thousand dollars we could do a focus group and get some real good answers. That's the way it continued for quite a number of years. I would do some speaking and lecturing about it, and at the same time I was teaching at a university — my specialty was American history — twice a week for 21 years. Until COVID came.

When COVID came, everything changed, as everybody knows. We were very limited in what we could do. I had cases in other states that refused to open their facilities to us. I finally got a group in Tennessee to let us plan a focus group with some modifications and masking, and then we did a massive case in Texas, and things began to loosen up a little. But the doors never reopened in a lot of places that were designed specifically for focus groups. It was getting very, very expensive to rent facilities — we'd have to go to a hotel, or find a facility used by commercial establishments for their own focus groups and convert it over. And I noticed that many of our clients were reluctant to re-engage because of the cost. Typically, in the metropolitan area where I was, the cost for a 15-person focus group could run as much as $30,000 or $40,000 for a day.

So then people came up with the idea of these online focus groups, where the respondents remain in their living rooms or bedrooms and participate. I went out and watched a few of them, and I thought they were awful. There was no way to control the attention span of the participants, no way to control the interference, no way to assure my clients that we were getting through to anybody. And of course, you never got that one-on-one rapport that I always loved to have with the respondents. So I thought that was a bad idea.

But when this all happened, I started to look around to see if there was anything viable, and somebody mentioned this whole idea of AI — which I knew nothing about. I started to do some research, and I found a couple of companies that were interested in developing this. This goes back about a year and a half to two years. I didn't feel comfortable that they could give me the kind of feedback I needed for my clients — because most of the cases I handled, on both the defense and the plaintiff side, were substantial cases involving a lot of money and some really significant fact issues, and I needed to feel comfortable that I was getting the results I needed.

And then I met a young man who was on the cusp of developing this platform, specially designed for litigators like me. Together we put together the information that we needed, and his tech team responded magnificently — to the point where I feel that what we're getting now is better than anything I could do live.

One attorney made a fair comment to me. He said, "I can't see their facial expressions; I can't see the tears in their eyes for my client." And I said — but the guy you're looking at in that focus group is not the guy who's going to be on your jury. So what does that matter? What we have to do is find a way to duplicate the discussion rather than the emotion. And that's what we've been doing.

Now, let me tell you what we can do. A client comes to us with a case. We talk to him about the case. He then provides us with deposition transcripts, expert reports, whatever court orders were involved, the complaint, the answer, the defense discovery, the plaintiff's discovery — and all of that gets submitted to our platform. We take in every document and every statement in the case. The synthetic jurors — and I'll use that term in quotes — are educated to analyze everything that comes in, and they feed back to us their thoughts about the issues: what are the most important issues, what are the highlights, what are the low sides. And we don't even get into a focus-group session yet — we're just analyzing the data. The program then produces a statement about the case, and we start from there.

The next thing we do is a survey of 130 to 135 community representatives, venue-specific. If the case is from the third district in Florida, then we get a demographically representative group of jurors for that venue, and we ask questions — a general community survey. After that's done, and after the attorneys have had a chance to review it and see where the important issues or the problems are, we narrow it down to a focus-group size — generally around 25 or 30 people. That's not something easily done in a live session. And we can do it within forty-eight hours of getting the material. We get back discussions — answers to all of the questions asked by the attorneys: what do you want to know, why do you want to know it, what's important, what's not.

The platform will also analyze deposition transcripts, videos, the diagnostic tools the doctors or the engineers have used. It will analyze the demonstrative evidence we submit — its value, and whether or not it supports an issue or a claim. And on top of that, we get conversations between the synthetic jurors as to why they voted one way or the other, and then we can ask follow-up questions on the information we need. That's, in general, how the whole thing works.

Collin McKee

Great backstory. So let's put it into context — you shared a story of a bicycle accident on our first call. Tell the audience how this software is actually catching things that a real jury pool might miss.

Jerry Goldstein

Well, that was one of the most interesting ones, because it was right at the beginning of my relationship with this company. This was in a major city with a big public transportation system — a big bus system. A woman was riding a bicycle in a designated bike lane. The street's makeup was: a sidewalk, a parking lane, then the bike lane, and then two traffic lanes. She was in the bike lane; the bus was in the right-hand traffic lane, several hundred feet behind her, and she knew it was back there.

As she was riding, there was a car parked in the bike lane ahead of her. She had two choices: go right, through the parked cars, onto the sidewalk — or ease out of the bike lane into the traffic lane. So she held up her hand and moved into the traffic lane slowly. The bus sounded its horn, and rather than slow down or stop, the driver tried to pass her on the left — and eventually squeezed her between the parked car in the bike lane and the bus. She fell and she was injured.

So the questions were: where's the liability, who bears the responsibility, how do we assess it — and finally, what are the damages? On damages, there were a number of treatments for different injuries, and there were doctors who said some of the conditions predated the accident, so we had to work around that. We represented the plaintiff at the time.

We submitted all of this to the panel, and the panel assessed a small amount of liability against the plaintiff — some of the synthetic jurors thought perhaps she should have attempted to go to the sidewalk side rather than out into the traffic lane. Then there was a division of responsibility between the guy who parked his car in the bike lane and the bus.

But here's what we found when we analyzed what the synthetic jurors did and said. First: they said this woman was Portuguese. They found that out from the depositions — by analyzing the deposition transcripts, they knew she was Portuguese. And since this was a demographically representative jury, there were people on it who were European — from Spain or Portugal — and one of the jurors recognized the fact that in Portugal, if the bus sounds its horn, that means it's okay to go into that lane of traffic. So whatever she thought was happening, she thought the bus was giving her time to get into the lane.

The most astounding thing was that the jury analyzed the video the bus had taken — I should explain, the bus was equipped with sixteen cameras, outside and inside, which is normal for a city that size, so we had a live video recording of the accident. And the jury analyzed the fact that the driver had, I think, nine seconds to slow the bus down or stop it — and didn't. He didn't apply the brakes at all. We didn't even realize that until the AI jury mentioned it. Then we went back into the depositions with the attorneys, and sure enough, it was in there. We hadn't realized that before we even started.

Then they analyzed all the medical records and decided there were a couple of items that may have predated the accident but for which she'd never received any treatment, so they disregarded a lot of that. What happened next: the attorneys took all of this to a mediation with the attorneys for the defendants and eventually settled the case for a very generous sum — matching the sum that our jury thought it was worth. It was a phenomenal exercise, and certainly an eye-opening education for me.

Now, one other thing I left out. When the program develops this demographically representative jury, the attorneys are given facts about every single juror: name, age, gender, race, religion possibly, employment, income level, marital status, children, military status, even political leanings — all kinds of information. So the attorneys can look at the jury makeup and say to me, "This came from XYZ County — perhaps we should look at a lower income group for that county, or a higher education level," and we can maneuver much the same as you would on voir dire and see if we get different results. And finally, we can reverse-engineer the whole thing — I'd say to the plaintiff, let's see what the defendants would learn if the defendants had hired us. So the information I get from this so far exceeds everything I would get in a live focus group.

Collin McKee

Yeah — just that cultural context about the Portuguese custom with the bus. If you hadn't known that custom, that context changes it completely.

Jerry Goldstein

It's custom, exactly. If you're near a bus today and the guy blows his horn, you're going to stay back in that bike lane. But the bus driver blew the horn, she waved, and she went into the traffic lane. That's what she thought she should do.

Collin McKee

That's what you do — especially where you come from. So have there been cases where the AI surfaced something that completely changed how the firm approached the entire case?

Jerry Goldstein

Yeah. We had a case up in — I won't tell you where; it was a northeastern state — and it involved long-running child sexual abuse at a school. I knew the attorney, and he called me and we talked about the case. Apparently the parents of this child became aware that the child was spending a lot of time with a person on staff at the school and never thought anything of it. When the young man went away on overnight trips — camping trips, one of these traveling teams — they never suspected anything and never did anything. Nothing about the relationship concerned them. It was a small town. And the attorney was worried that the jury might punish the parents through the child, because the parents didn't do anything.

So we took the case and put it through a test program we had. And after I analyzed it with him, I said: you're approaching this the wrong way. The parents are not parties to the case — don't even talk about what the parents did or didn't do. Why don't you focus on what the teachers in that school saw and did nothing about? They knew this guy's behavior was improper. His door was closed when he was meeting with the child. The child met with him almost every day of the week. The child rode in his car leaving school or arriving at school — all of which was improper, and all of which should have raised red flags and bells and whistles, and didn't.

So I said to the attorney, let's look at what the jury says — and that's exactly what happened. When they were asked the right questions, they came back and said the concentration has to be on the school, on the staff — the school never followed the cautionary requirements they were obligated to follow. By the time this case goes to trial, the victim is now an adult; and under the laws passed in that particular state, the statute of limitations was practically thrown out, so there was no limitations problem. The case isn't over yet, so I don't know the result — but I'm sure the advice we got from the program is going to change it. The attorney was astounded that he had missed this entire direction. He was so worried about what the jury would think of the parents that he missed all the signs and signals that were evident at the school district.

I always used to tell attorneys: look, you've been living with this case for the last three or four years. Give it to me for an afternoon and let me put some new eyes on it. I can pick up stuff that maybe you missed — the forest for the trees. And that works. That's exactly what I think we're doing with this program.

Collin McKee

It's obvious that it's working. So why do you think it's been slow to be adopted?

Jerry Goldstein

There's a very simple answer for that. Attorneys are the worst people — and I'm speaking from my own personal experience — the worst people to get to change, to try something new. There was an article I read years and years ago in Trial magazine, where an attorney wrote: if you've got a case that's worth a hundred thousand dollars and you don't do a focus group, you're almost at the level of malpractice. And I believed it — I really do. Now, that was a long time ago; $100,000 was a lot more money then than it is now. But if you've got a substantial case, and you've got an issue or two where this can help win the case — for a defendant or a plaintiff — you owe it to the client to do this.

And now, with the way we're doing it with AI, the cost is a fraction of what it would be. You don't have to leave your office, and you get the answers within a week. If you came to me today — today is Monday — and gave me all the material, by next Wednesday we'll have a report for you, or sooner if we have to. So not doing it is just plain wrong, I think. And then you get the guy who says, "Well, I've been trying cases for 30 years." Okay — so what? I've been trying cases for years too, and I know how to try a case. That's fine. But sometimes you miss things.

So: the cost is now a fraction of a live group, you're getting more information, more opportunities to analyze the case, and you never have to leave your office. What could be better than that? But attorneys are just so slow to move and change. The other thing is that attorneys are using some of the commercial platforms that are out there — Microsoft has something, and there are a couple of others. First of all, they're not confidential, and the attorneys who use them run the risk of having to produce those results in discovery — the courts have recently said that. So that's another problem. And the other problem is that none of them have the advances that we have on the technical end of what we do.

Collin McKee

So what's working as far as bringing it to market — and what isn't? You obviously have clients. Are they coming?

Jerry Goldstein

Word of mouth. And the other thing is that the insurance companies have found out about us, and they think this is the best thing since sliced bread, because they're saving hundreds of millions of dollars in legal fees. That's really what's driving the market. So I say to the attorneys I engage: all we're doing is balancing the playing field. We don't make a choice — if we get called by an insurance company, we get hired by them; if we get called by the plaintiff first, then we won't take the defense side of that case. So I'm saying to the plaintiff's bar: you've got an opportunity to balance out the playing field here for a minimal cost. And we're getting some responses, but not as many as I would think.

I don't know how best to market it other than what I'm doing with you today — and inviting people to contact us. If you're an attorney, I'd say: send me a letter with just a few paragraphs about the case. Tell me what the issues are, what the damages are generally, something about your client. Give me that letter, and in 24 hours I will have a preliminary report for you — all about the case, just from those few paragraphs. The attorneys who have done that have been astounded at what we can do with just those bits of information. And then I say: imagine what I can do if you give me the whole file.

And for the larger firms — we deal with firms that have anywhere from ten attorneys to hundreds — what we're telling them is: we can set the program up within your computer system, so it's inside your system. We'll teach you how to use it, you get the same information, and you just pay for the time you use the program. There's an upcharge for it, but the defense firms especially love it. Those are the kinds of things we're working on — convincing people that it works.

Collin McKee

That's excellent. And you said that's been just the last two years. So where do you see this going in the next two years?

Jerry Goldstein

I don't know where the end is. I don't know what more we can do. The attorneys ask us, how do you educate these synthetic jurors? I'm not a technical person — I just talk to my guys and say, this is what we need; this is the information we need.

Now, interestingly, I just got a call on a big case where I found out yesterday, doing some review of the documents, that the case has been bifurcated between liability and damages. I don't know if you're familiar with that concept — the judge says, this is a very complicated, massive case, so the jury is first going to hear the case just on liability: did the defendant do something wrong, did they cause the problem? If the answer is no, the case is over. If the answer is yes — the defendant bears some responsibility for what happened; it's a commercial case, by the way — then we get a whole new trial on just the damages. So we're trying to figure out the best way to handle that with AI and see what we can do. If we can, we can program it that way and see what happens. That's what we're working on right now.

Collin McKee

Excellent. Jerry, this has been a great discussion — I think it's going to be eye-opening for a lot of folks. Hopefully our audience will reach out to you. What's the best way to get in contact with you and your company?

Jerry Goldstein

The company website is njtrialconsulting.org — NJ as in New Jersey, where we started. And my email is jerryg@njtrialconsulting.org. And I'd be glad to answer a couple of questions — if somebody just wants to chat about a case, I'd be more than happy to work with them.

And by the way, it doesn't matter what jurisdiction you're in. Our program is designed to meet all the jurisdictional requirements in terms of jury selection and venue selection. Right now we've got a case in a federal court that encompasses several different counties, so the jury has to be selected from those counties — and we make those adjustments as well.

One other thing we do — can I take a few more minutes? We've found big law firms that do a lot of arbitration, both nationally and internationally. We now have a program where we can gather every bit of information possible on the arbitrators, so that we can assist the attorneys in making their arguments to those arbitrators — with a great deal of predictability in terms of what the arbitrator would do with the specific issues. The people who engage in those kinds of cases are just thrilled with what we can do, whether it's a judge, a mediator, or especially arbitration. And in international arbitration, everything is done on paper — there are no witnesses — so we're able to do a great job with that. That's working out well as well.

Collin McKee

Awesome. Well, Jerry Goldstein, thank you very much for this very entertaining and educational conversation — really appreciate it. And if there's anyone in the audience who's interested in their readiness with AI, that's what Endeavors AI does — we like to audit workflows and check into where additional efficiencies can be found. Links to both Jerry's company and Endeavors AI will be in the show notes. Really appreciate you watching, and for Endeavors AI and Jerry Goldstein — we look forward to seeing you in the next episode. Thanks, Jerry.

Jerry Goldstein

Thank you.