RISE AI, Part 3: Shaping the Future of Voting with Technology Innovations

Episode Topic: Shaping the Future of Voting with Technology Innovations

The intersection of Artificial Intelligence, democratic integrity, and election security represents the primary frontier for modern governance. Dr. Juan Gilbert’s pioneering research establishes a strategic benchmark in this space, transitioning from the historically reactive "hanging chad" era toward a proactive technological paradigm. By integrating universal accessibility with architectural security, Dr. Gilbert addresses the critical vulnerabilities that compromise public trust and electoral resilience.

Featured Speakers:

• Dr. Juan E. Gilbert, Andrew Banks Family Preeminence Endowed Professor and Department Chair, Computer & Information Science & Engineering, University of Florida

Read this episode's recap over on the University of Notre Dame's open online learning community platform, ThinkND: https://go.nd.edu/e2b163.

This podcast is a part of the ThinkND Series titled RISE AI. (https://go.nd.edu/32b04c)

Thanks for listening! The ThinkND Podcast is brought to you by ThinkND, the University of Notre Dame's online learning community. We connect you with videos, podcasts, articles, courses, and other resources to inspire minds and spark conversations on topics that matter to you — everything from faith and politics, to science, technology, and your career.

• Learn more about ThinkND and register for upcoming live events at think.nd.edu.
• Join our LinkedIn community for updates, episode clips, and more.

Transcript

Welcome and Opening Remarks

Speaker 10 0:06

Hi friends. So it's, so let's, get started with the, uh, next keynote so we can stay on time and track and we'll allow people to stream in as, as they come in. But, we have heard some amazing con, we had some amazing conversations, dialogues ranging from, you know, since yesterday about, AI and, and then for Latin America, large language models for all. we've had some of the former PhD students of Notre Dame talking about where they are now. They graduated about 10 or 15 years from, uh, now, uh, ago, and it was such a gift of their time to come back to the alma mater, about 10 of them, and share with you where they are, what they are doing, and where they are headed. Then we had an amazing, uh. Evening of, uh, opening ceremony with, with some of the accomp, some great, uh, and leaders amongst us sharing and, uh, sharing their findings, followed by the Keynote by Ascension. So there's a lot that happened yesterday itself. And then the morning we kicked off with an awesome paired keynote by, uh, Accenture, where, uh, both Dale and Ana really hit home. A point I thought was very fascinating, where, start with. The business realization of ai. Don't just get into AI for the sake of doing ai unless you can attach it to a real business outcome. Start from there and then build up your data stack and responsible AI gets woven in and how you operationalize gets woven in. So it was, that was very interesting. And then we had a panel discussion on AI and sustainability and parallel tracks on healthcare. And now we are all gathered together again to have, uh, you know, one of my good friends and, and real, a, a a distinctive academician and a world renowned scholar who's gonna tell, talk to us about, AI voting and democracy. And, and this is an issue that at Notre Dame, we think about, we have a strategic initiative on data, AI and computing as I shared yesterday, as well as an initiative on democracy. So, um, I'd like to welcome. Dr. Hanani Gilbert from the And who's the Andrew Banks family, preeminence endowed professor, the University of Florida Distinguished professor and also the Chair of Computer Information Science and Engineering Department at the University of Florida. We will have to ask him how he does it. All right. and he directs the impactful computing for Social Good Lab at Florida, and he is one of the most highly decorated computer scientists in the nation. His influence has been recognized. At the highest level. He was named the Laureate of the National Medal of Technology and Innovation by President Biden, who also appointed him to the prestigious National Science Board in 2024. Uh, he's a fellow of, uh, association for Computing Machinery, a CMA fellow of, and a member of the American Academy of Arts and Sciences, a fellow of the National Academy of Inventors, among many, many honors So, so Huan. It's definitely a joy. To have you here. And he also has several friends here at Notre Dame, including I'm looking at Ron. so he was delighted to join us. I, at first thought he's delighted to join us for Rise ai, but then I found he's actually more looking forward to seeing his buddy Ron here as well. So, but, but thank you so much Ron for, uh, for joining us, taking our time. And, and on a CMI would also like to make an announcement. Uh, we have Sean Pigeon Herero, who's the executive, uh, publisher with a CM, all the abstracts that have been accepted to the conference. And a CM has agreed that we will have a rolling, rise AI editors collection at a CM AI letters. So if you submitted your abstracts, and we will invite you, the program chairs will invite you to submit your uh, uh, papers. To the, uh, AI letters journal that a CM has launched and, uh, and, and no bias. I'm one of the editors in chief of that. but you will have the RISE AI rolling, uh, collection for a CM AI letters. I'm really pleased to announce and thank you, Sean, for making the tip to joining us with that. Juan, please. Thank you.

Speaker 2 4:10

Thank you. so. I've been working in this field since the year 2000 and working with elections technology, and today I am going to actually reveal for the first time public, a new innovation coming out of the lab. Uh, whenever I talk about these things, it gets very engaging. I welcome interaction. all questions and comments are welcome. so don't hesitate. Okay.'cause we, we learn from that. That's, healthy to have those kind of discussions. And I think sometimes, um, the unexpected is, is where we find those, those little gold nuggets of good innovation. Okay. Without further ado, so what I'm gonna do is tell you about shaping the future of voting technology. Innovations that have come out of the lab and how we've, we've actually impacted voting in particular in the United States and that I'm not, uh, exaggerating. We've changed voting in this country through our lab, and you'll see that. So some of the things I'm gonna talk about, um, as we see AI and technology becoming. More prevalent, pervasive in society. What is the future of technology in voting in general? will people ever be able to vote online? And if not, why? Uh, what about the long lines for voting? Can technology eliminate them? I wanna talk about that and maybe get into it, but I don't know if I have a lot of time, but I'll make, I can make a comment on that. And then voter id, that's been an issue, uh, of late in this country in particular. And can technology and AI help? again, I, we'll see how this goes because it can get very conversational. I, I, I want to say it again. I welcome the questions in the conversation. Okay. Alright. So let's take a look at some technology that come out of the lab, and discuss the questions along the way. And so I'll begin with just a little background. Currently state of the art are, are what's called direct recording machines, DRES. These are machines that are basically computers with a touchscreen, allows you to touch it, you mark your ballot. And, um, there've been some issues with these lot of politics around them. I won't go into detail. You could see some of the headlines there, but, uh, ballot design is one of those issues and, um, and accessibility, but we, we'll talk a little bit more about that. Background next, uh, optical scan. optical scan is something we, we, we use in, in the country and you see it around the world. Uh, essentially you have a paper ballot, you mark it, you scan in an optical scanner. Now this technology is, seems, uh, thought to have been very secure, but it has some issues recognizing stray marks. subjectivity in recounts and did it count your vote as you recorded it? And then there's some issues around how they, the number of ballots that are actually scanned or not scanned. So, little background, Minnesota, when we talk about optical scans, I always go back to 2008 because optical scans were thought to be that's the way we should vote. We had a huge issue in Minnesota in 2008, and I'm just gonna demonstrate the issue. Here's an optical scan ballot. These are real ballots by the way. You look at this ballot and you say, well, who do this person vote for? The optical scan would say there was no vote, but a human looks at it and say, oh, they won. Norm Coleman. I don't know. What about this? Looks like they wrote the word yes in the oval or something. Will the optical scan, pick that

Speaker 4 8:08

up. self-explanatory. And here's one.

Speaker 2 8:19

So, so you look at these ballots, could, how they interpreting the scanner, if we applied AI behind this, could AI even interpret these? And how would it, what? What model would you feed it to? Learn how to do these kind of things? Humans look at this and you could debate who they want to vote for. This is, this is reality. And again, these are real ballots. I didn't make these up. These were actual ballots that were taking from that election. So with that introduction, let me tell you about what we did. We've created a technology called Prime Three stands for Premier third generation voting system. When we created it, it's open source. The idea was to create a voting machine that was usable, accessible, and secure. And it was one machine that everyone could vote on, and we designed it with a universal design in mind. One machine, everybody votes on that machine, independent of your ability or disability. So we made it where it had touch capability and we actually put voice in it where you could speak to it Now. Speaking to, it was not saying who you would vote for. So in the previous example, you wouldn't say, Al Franken, it would say The vote for Al Franken say, vote, and you can respond by saying, vote or blow into the microphone. We did that because guess what happens now? I'm immediately language independent. So with this technology, we didn't realize it, but if you can't see, if you can't hear, if you can't read, if you don't have arms, you could vote on the same machine as anyone else. This came with a extra benefit because if everyone's voting on the same machine, it had a security benefit, which is that if I'm gonna deceive you not to worry about deceiving people with other abilities or disabilities as well. So from a security perspective. When we designed it, we actually went old school and we designed this machine. It's a stateless machine. It has no hard drive. It runs off Bootable Blu-ray read only, so you write it, it boots and it cannot be written to again. And I actually had the opportunity to talk to colleagues in security who worked in voting. Who had demonstrated hacks of voting machines and I offered to send them the machine and said, I want you to hack it. So they would flip votes. And then I wanna see if we can actually, if it persists and if you can send it to me and if we can find it. Date no one's taken my offer. What this machine does, it prints a voter verified paper trial paper ballot, voter verified ballot. So you mark your ballot and then it prints, and here's a sample, of that ballot. I'm just showing you what's the nist, uh, national Institute, standards and Technology. Uh, they create a standard ballot for testing voting systems, and this is it. Now, if you notice each contest, it only prints the contest and who you selected, and I have a picture there because we actually could put pictures on the ballot, which were helping in an accessibility way as well. By printing this way, it eliminated ambiguity about whom you voted for. And, uh, it was more precise. So you print the ballot, that ballot goes in a ballot box, and then we had a scanner, which we would scan it and we scan it using optical character recognition. But we added, our own spin on that and it's, we call it IOCR, informed optical character recognition. so think of it like, you know, in ai we use a language model to build the actual, the model. We, we use the ballot somewhat as a, as a language model to build the OCR recognition. Yes.

Speaker 10 12:24

Ask.

Speaker 2 12:25

Yeah, go ahead.

Speaker 10 12:25

So, uh, the, so your machine, your system, it's only a voting mechanism. I mean, it doesn't store any data about the world. If you get the ballot out, you independently verify and it goes to the old

Speaker 2 12:38

absolutely.

Speaker 10 12:40

So you're not changing the counting process, you're changing the voting mechanic.

Real-World Applications and Case Studies

Transparent Voting Machine: Addressing Vote Flipping

Speaker 2 12:44

We changed the voting mechanism and right, and we changed the way the v votes are tallied with this optical character IOCR. And then, there was this technology risk limiting audit is a, a stark invented this. It's a way to confirm that it is actually the machine is accurate. So we created this and our IOCR is extremely accurate. we tested this and that is part of our process. So yes, essentially, not essentially, but you're right. The voting machine has no way to store anything. Again, it's stateless, it has no hard drive. Yes. The ballot definition and the actual software, the operating system, we could tech out if you want. Uh, it's, it's a Linux bootable, uh, this, and it's running just our software boots up and allows you to whoop and prints. Now I'll show you a little bit more about the system in a second. Okay. So with this technology, state of Wisconsin actually used it back in 2014. Uh, New Hampshire used it, in 2014. This was their setup. Actually. They had, as you see in this, picture here, let's see, we have two. So you see a headphone, a tablet, the printer. And this thing here is a button switch, two button switch on the right side of it. And the the left side has texture, the right side is clear. So if I am a blind voter, I put on this headset and I use the clearer, it says Next I go to the next thing, next thing, next thing. And I hit the texture to make a selection and it would print. And, so they use it. So they actually created, and they called it one for all. They renamed this system. So they had statewide use in 2016. Ohio used a version of Prime three, but they're using it as their remote, accessible absentee system. So what they did was, in particular for blind voters, for example. They would have to register with the election office. They would actually get an email with a, with a link to the ballot where they can mark it, print it, and mail it back in. I mean, it's cumbersome, but it works. ESNS is the nation's largest voting machine manufacturer. This is their technology called Express Vote, and this technology implements. A lot of the things that we created in Prime three, it's not stateless to my knowledge. prints a paper ballot. I don't like that. They, they, they went a receipt side. That ballot, you see, that's too small. I, I, I'm not a fan of that. but they did implement, a, a cool aspect. One of the things we did in Prime three was that you could go online, fill out your ballot, and generate a QR code. On election day, walk up to the machine, scan it and it takes you into the review mode. You review everything and you can print so it sped up voting. They actually have that in that machine. I don't know if they advertise that though. I had the opportunity to serve on a national academies, consensus committee securing the vote, protecting American democracy. We published this report back in 2018. The report strongly encourages audits and paper ballots, and we had a debate come up where hand marked paper ballots versus the ballot marking, device, but we said, you gotta have a paper ballot. Why? There is no known way given the current state of technology to secure a digital ballot. I'm gonna say it again. There is no known way given the current state of technology to secure a digital ballot. I don't care if you're encrypted or whatever, whatever, you can't secure it. I can manipulate it one way or another. And, and the manipulation. So most people think, well, if it's encrypted, how are you gonna change it from, uh, Biden or Trump? It's not necessarily that I gotta change it to Biden or Trump. I can contaminate it so it can't be tallied. That's an effective attack. You'll see more about that kind of attack in a minute on paper ballots. So we gotta have a paper ballot of some sort. The problem with the paper ballot is it's not accessible. If I'm blind, I can't manipulate a paper ballot without help. But then that says, well, we have to use a ballot marketing device. But then the security. Folks say, well, I can hack it and you know what I can do with that this machine, I'm gonna flip votes and you'll see what we that means in a second. So I pointed out issues with hand marked paper ballots, and I'm gonna show you some that no one talks about these'cause I guess it's not, it's kind of depressing. But, but I will share it with you. under Vote Hack Over Vote Hack, and a Stray Marks hack. This is the Undervote hack. Often people say, I'm not gonna vote in that contest. I don't really care about them. So they don't mark anything. An insider afterwards could look at that and say, Ooh, I'm gonna vote who I want. Takes less than three seconds. Impossible to detect. But if you don't mark something, boom, someone could come in and mark the candidate of their choice. Impossible to detect unless you catch'em. Doing it in the act over vote. Well, this person voted for norm, but I really want al to win and I'm an insider. You know what I'm gonna do? I'm gonna mark al that causes an over vote scenario. So guess what happens? Ballot is that vote, that contest is thrown out. Norm had a vote, but I just took it away. That's a over vote hack. Again, less than three seconds. Impossible to detect. This is what happens or can happen. So, and we already saw the stray marks, from the Minnesota where people just write on the ballot. So issues with ballot marking devices. They were using barcodes or QR code and they were saying, well, what if the ballot text doesn't match the actual QR code that's being scanned? And then the vote flipping, that was a big deal. Would voters notice if, I, on my ballot marketing device, I mark it, but when it prints, it prints something else. So University of Michigan did a study. And, you'll see it here in, in their study without intervention, only 40% of the participants reviewed their printed ballot at all, and only 6.6% total poll worker, something was wrong. So what they did was they held a study, let people use a ballot marketing device, and when it printed their ballot, it flipped their vote and they wanted to see if people would notice. And this was the result when this paper was published. It sent the country into a shock election community. They said people won't notice. So Rice University did a study. They said, well, we did the study and our data suggests that voters can detect changes if they only attempt to do so. Well in their study, 76% detected anomalies indicated that voters can reliably detect it, but if they attempt to do so, yes. Well it, from a statistical standpoint, it seems low, but from a human perspective, it's extremely high. And I'll tell you why. How many people does it take to show that the election is, has an issue? That number's very low. And I, I'll, I'll come, come back and comment on that in a second. So, rice also said, while the overall detection rate in our study experiment was 17.6%, but they noticed 76 could detect again. They said, if they only try, so we that, we said, all right, we're gonna try and fix this. We created what's called the transparent voting machine, and again, it's powered by Prime three. So take a look. So what you're looking at is the trans, the the near drill prototype. And you can see I'm marking a ballot and it prints right in front of me behind that screen. In order for me to continue, I actually have to touch what was just printed. Essentially I'm attempting to turn paper into a touchscreen, so I can't continue unless I touch what was printed. If I touch outside of that, it will actually tell me I have to, you know, touch my selection. That's what I'm doing here. and again, this is, this video is from the original prototype. It's a little different now. You'll see another version of it. And I'm just writing in a candidate and doing some things here. but the transparent voting machine was an idea, and we made it a reality. And so the idea that we, the problem we were trying to address is, again, remember vote flipping. You saw the Michigan study, the rice study. People were very alarmed by that. And so we felt that if we designed something like this. Would this make a, an improvement? Alright, so again, uh, I'll let this one print and then I'll move on. And again, this would run in a stateless environment, so I don't know how you get it to flip boats, but just assuming you could, I didn't mention this, the stateless machine, our proposal. That the stateless machine, physical machine would be designed with like this, see through, and I would use AI to actually image the before image of it, and someone made an accusation. I could then use AI to look at the image of the inside of the machine to see if anything was changed. Yes.

Speaker 6 24:11

So how long did it take for you to be in this like voting books for, have all this done?

Speaker 2 24:17

It did not. The amount of time it takes to vote this way is same as paper. We, we actually have the data on that. It's, it's a lot. It's, it looks like a lot and, and it's gonna feel like a lot in the videos. Remember, I'm shooting a video to demonstrate, so I'm actually moving slower than what would normally be. And again, remember we have the opportunity to scan the go online, mark a QR code and scan it, and then just go straight review. Alright, so the transparent voting machine is using our open source prime three. That was the prototype. We published a paper about it, but will it enable people to detect flipping boats? We did a study, had 151 participants. I did this study coming out of COVID. Here's a picture of me, conducting the study. There's a participant, sitting there and, in front of the machine and I am there at my, at a laptop. And what I'm doing is observing everything that she's doing and recording everything that's happening. Okay, so we did this study and what happened? Drum roll. More than 40 of the participants noticed to flip, but didn't say anything. How do I know that? Look at me. This is what happened. They were voting I I flip a vote and they do that. They didn't say anything. So I noticed a physical reaction to the vote. Flipping what that is, some of you may be familiar with this, called the Hawthorne Effect. When you do a study, you participate in a study and you know it's a study. You behave differently than you would in reality. That's what they were doing. It was a study, eh, I don't care. So what we did was I added a question at the end. Did you notice I changed one of your votes to determine if they actually noticed and they just wanted speaking up. Well, 6.6% detection rate in the Michigan study. At best, 40% reviewed the ballot, 17.6% overall in RICE study. Okay. And 76% could detect anomalies. We had a 77% minimum overall detection rate, and only coincidentally 6.6% did not notice and could not identify the flip. Of those that did not notice, 71% correctly identified the change. When we let'em know it happened, they, they, they, they, and in the review they could identify it. So overall, I had a 93% correctly identified the change given the overall minimum notice rate of 77%. And that excludes the on paper notices. It is unlikely an election could be hacked such that a hack could go unnoticed. So here's that number. Let me tell you why that matters. Here's what we noticed. The ballot I was using was the, the previous 2020 president election ballot, okay? And what I noticed was if I flipped a vote from, from Trump to Biden, for example, I had a few people wanna cuss me out. So what I noticed is that if you flip the wrong thing. All it takes is one, and you get this passionate response. The power of one is big in this area as far as if, if something happens. So you don't need big numbers to, to identify something's wrong. We published a paper about it in 2021. the Harthorne effect was important to notice. Neither Michigan nor Rice accounted for that. And it was just something I observed in collecting data, a physical response to a vote flip where they didn't speak up. So we don't know how many actually may have had that in Michigan and Rice, we don't know. Yes,

Speaker 7 28:29

presidential was there a, we noticed that they had much more vocal.

Speaker 2 28:38

I have the data. We didn't document it that way, but they, it was only that contest that they really spoke of, and you can imagine why, right? So, but yeah, I, they noticed with this. So what did we ended up doing? We have another version. This is the next version of it. it's not the big box, it's slim. Uh, it prints, and here's an example if you don't touch it. So we created a transparent interactive printing interface, um, which is a new technology in itself. Has anyone adopted it? Not yet. I don't know if Tim made it. Is Tim here? No. Oh, anyway. This is a, a technology. We, we patented the technology. University of Florida owns the rights to it. We'll see where it goes, but we know we can securely get votes this way. Yes.

Speaker 3 29:46

But second of all, um, I'm curious about that selection process. What,

Speaker 8 29:50

you know, if you build a better mousetrap here in the, in the voting ecosystem, what's the process of they selling the states? Who makes those decisions? And then how are they becoming aware of this and making it,

Speaker 2 30:02

they're becoming aware of our technology just by, I'm, I'm fortunate to have a profile with the vendors. People know about us, and we're promoting this now. I think, I'm, I'm, I apologize, but I'm just sharing the reality. This works, but it won't be adopted until we have a next problem. I'm sorry, that's just, they're just responsive to problems. They just, I'm sorry. So I don't know that we'll have that anytime soon. Yes.

Speaker 9 30:34

You, you made a claim and you've emphasized it a couple of times, that there's no solution for error digital ruling. and of course the first thing that comes to mind is, is having private key, individual signature and then you make complain. Well, you can, you can still sort of disable those folks by manipulating it seems like there's an analog to the self-driving car situation. It's not that the self-driving car needs to be 100% safe, just at least the savings of human driver. So you sort of went on to, to mention one of the issues with paper ballots and how systemically that can be addressed. To the extent that paper ballots themselves are not possible before.

Speaker 2 31:16

That is a great question. And the answer is, it's a so answer. It's not a embedded in data or anything. It's the, the psychology, sociology of voting politics, to be honest. That's, that's really what it's.

Speaker 4 31:44

So how much of your results been told that the choices

Introduction to Tele-Voting

Speaker 2 31:51

well down and, and we tested this with a real ballot and down ballot with things that people didn't know about and we flipped and they still saw it. Because remember, what I'm forcing you to do is you are touching something that just printed, that you just printed and, and you can't see it here, but, let me see. Yeah, I don't have a picture of one, but, but essentially what happens, you see the box that's around what's printed, so that would have president and vice president Arrow in your selection. You see the text above it. Before that box comes up, it would say, actually, you see it in the video. I'll show a video in a second, but it will actually have the name of your selection as well. So your selection appears on the screen and in the paper as well. I do that deliberate, that's a security feature. If you're gonna hack the system, which one you gonna flip? If I flip both of them, you just increase the likelihood of me seeing it by flip. One, you're providing evidence that a, a hack has occurred. So, yeah, it, it's, for us, what we've seen is if it doesn't matter to you, you still see it. That's the point. Alright, lemme get to the, the, the new stuff. Alright. Tell a voting. So can we vote on the internet? the problem again, resulting in a paper ballot. The current standard of practice is. Like I mentioned in Ohio and some other states, I don't know if Indiana has this, but, uh, remote accessible, voting or ballot marking. You mark your ballot, you print it, you mail it in like absentee. So we had this other idea use using the transparent interactive printing interface as well. You're the first to see this publicly.

Speaker 10 33:45

Mute audio button. NIST Chrome, NIST Standard ballot. Google Chrome. Heading level two. To start voting, press the start button below. Start voting button. Heading level one, US President and vice president. Heading level two. Vote for no more than one candidate. Heading level three, contest one of seven. Joseph Barky and Joseph Haller and Blue Party Adam Kramer and Greg Volo Yellow party. Daniel Court and Amy Lombard Purple party, Alvin Boone and James Leann Orange party, Alvin Boone and James Leann Orange party selected Austin Mar Char Marzi candidate reprint my selection. Alvin Boone and James Leen, US President and Vice President for Frame two, your selection. Alvin Boone and James Leen for us President and vice President has printed, press enter to continue heading level one US Senate. Heading Level one, heading Level two. Vote for no more than one candidate that the screen reader blind. You are currently on print, my Leaving Frame, US Senate vertical entering Frame. Your selection. None of the above for US Senate has printed. Press enter to continue. Heading level one US Representative Heading level two Vote group and heading level one. Go back. Heading level two, type in the candidate's name. Edit text blank. B-U-G-S-B-U-N-Y. Heading level bug. Heading level Bug bunny. You are currently on a button. Print my selection. Bugs Bunny button, US Representative vertical line frame two. Your selection Bugs Bunny for us Representative has printed, printed valid image heading Charlene Franz Blue party button Heading level one. Governor heading level two. Vote for no more than one candidate. Barbara Williams, Aqua parties. Print, Mica Print myself. Leaving frame. Governor, vertical line entering. Frame your selection. Barbara Williams. For governor has printed press enter to heading level one. Lieutenant Governor print, print my leaving frame. Lieutenant Governor entering frame You did not make a selection for Lieutenant Governor. Press enter to continue and heading level one County commissioners. You are current. More cap previous candidates. Helen Moore. Helen Moore. Orange party Select Valerie Alman. Orange Party pre print. My leaving Frame. County commissioners entering. Frame your selection. Valerie Alman. Helen Moore for county commissioners has printed. Press enter to continue. Heading level one, proposition number one. Heading level three, contest seven of seven. Retain Robert Jones', chief Justice of the Supreme Court. Yes, but no button, no selected. You are currently on a button. Unlabeled image? Yes. Selected print? No. Print my selection. Yes. Button. Leaving Frame. Proposition number one. Entering. Frame your selection. Yes. For proposition number one has printed, press enter to continue. Link. If leaving. Frame. Entering frame frame two. You are currently on a frame entering. Frame. Heading level one. Do you want to verify your paper? Paper ballot scan. Yes. Verify my ballot. No, submit my ballot button. You System, system settings. Voiceover of,

Future of Voting Technology

Voter Pass and Line Management

Q&A Session

Speaker 2 37:02

alright, so what you just saw there was someone voting remotely and their ballot printing in the precinct and them verifying that in real time. And after this happens, that ballot, if I want to verify it. Decide if they were forced to do it, but it's scanned in using our IOCR, and then a text is sent to them where they would get that scanned image, the physical scan image, and be able to verify it and accept it on a separate device, their phone. So this technology, you just saw the demo, it is working. And from a security perspective, it has all kinds of benefits. We could talk about. That is a mimic of the transparent voting machine, so that's why I was saying to you, you see how the name appeared and then it prints in the lower area as well. Here's, I've had opportunity to demo this to some accessibility communities and they like this version, but you'll see I won't play audio for this, but I'll just let you see what, I call this the basic version. Basically what it did it instead of, verify, I think the thought was that if I verify after each one, they thought that could take long, longer time and I could have an error or something. This one allows you to mark the ballot fully. You can navigate all over the place and mark your ballot fully and then review it at the end and you'll see what happens when it prints. it uses the same printing procedure, but just a different marking. So in this scenario, again, a blind voter or sighted voter, you navigate, you put in your, your votes, you can check'em over and over, and you'll see a review area here in a second. So I'm just able to play as long as I want and do what I want. So I'm gonna vote. Yeah, so go back. I change my mind. I change something, then I go to review. So these buttons allow me to go to any contest if I want and change it. But watch, watch what happens here. I print my ballot. Now what happens? You enter to the same interaction now where prints now gotta confirm it one by one. So it's gonna print on one by one. And I confirm we haven't studied this yet. Uh, obviously we studied the first one in the transparent, voting machine, but again. Talk to a group of accessibility advocates, and they seem to like this, so I put up a test site, so they're playing with them. We'll see what happens. Either way, the, the idea here is that, we essentially, we've created a voter verify paper ballot over the internet. That's what we've done, so. Going back to original questions, as technology becomes more pervasive, what is the future of technology and voting? I just showed you part of that. I don't know how long it's gonna take. Getting at your question. Usually when there's an issue, things change, but until then they don't change. Really. Will people be able to vote online? Well, I kind of explained that and gave you a demonstration of what we think could be done. Now that model of, of tele voting, which is what we call tele voting, it does require, uh, human resources on the back end, which I say is a good thing, as well. But I don't know, we'll see how far that goes. Well, people deal with the long lines. Well, we created another technology to address this. Which is, is kinda being in Florida, being down the road from, uh, Orlando and Disney. Uh, imagine you, you, you know, you, you can actually go to vote and we call this thing, we create a technology called Voter Pass, where you could pre as a, instead of getting a line, you print a pass, which is your, spot in a line, and it tells you what time to come back to vote and things like that. Now. That seems kind of like obvious, but it has a very negative side effect to it that election officials don't really like. In order for us to do the, to build voter pass, meaning we can do line management and print, these, these paper ballots or not ballots, these passes for you. We gotta do a queuing model. If anybody's familiar with that, what that tells me is I know the upper bounds for it, meaning I have X number of machines and if I've reached certain capacity, it's impossible for me to serve everyone today. They don't know that. But using voter Pass, we would know exactly what the bounds are. So some election precincts. Don't like that'cause they suspect they're at the bounds, but they don't want people to know they're at the bounds. So I don't know where, how far this would go, but we can do line management and make it work. Voter ID controversial, we came up with an idea. I think one of the things we should. We, we haven't tested this yet, but here's a proposal. Use fo facial recognition for voter verification. So, Naess, you know, Damon does work in this area. We kind of talked about this, it's voter verification. So in order to vote, I come up just like at the airport, and now I, I look into a machine and I'll say my name and, and maybe my address, or just say my name. So Na Natasha, if I wanna steal your identity, I gotta look like you sound like you know your name as well. Those three parameters would increase the accuracy of verification very strongly. So I don't know how far we'll go with this, but we've, we haven't implemented a system, we just proposed this approach as something to test, as a way to verify, because right now, IDs. IDs are controversial in elections. People are debating voter id and so we think this could be something we could test. I dunno. So lemme stop there and take questions. again, thank you for having me. this work was supported by all the people you see on the screen. Um. We're still going, so I'll take questions.

Speaker 10 44:48

Please use the two mics there for any questions you may have for one.

Speaker 11 45:00

Hi, thanks for that. Super interesting. my question is actually about the age of of voters, which you already, you already know what I'm gonna ask. Third of voters are over 65. Two thirds of voters are 45 and up in your tests. How did that, uh, kind of agility and ability to engage, whether it's online or on with the machine impact their uptick uptake of this?

Speaker 2 45:23

We haven't tested online yet. We're probably gonna do a spring study on the tele voting. We haven't done it, but the transparent voting interface, we did go out into the community and test the, the, what I showed you is just a demonstration of me. I was in the library on, on campus, but I did go out into the community and test, and we had older adults and they were able to use it successfully. It, it was pretty straightforward. The, the thing about the interface, uh, my students, it's funny because I'll show a demonstration of it. And, and people say, well, what's the big deal? I don't, I don't get it. That's the response we want.

Speaker 4 46:00

Mm-hmm.

Speaker 2 46:00

It's so easy that people think they could have designed it, and they think it's just so transparent. That's that's the whole point. It's not supposed to be, wow, that thing is so cool. It's supposed to be, what's the big deal? And yes, older adults were able to use it. Interesting. They, they've had touchscreen experience already. So it was a natural thing. 60. Well, in, in our studies, we were, we, we were, we had, I can't remember who the oldest was, but we had 65, I wanna say seventies. So we, we, we had a good demographic.

Speaker 11 46:42

So you won't, you won't need, you don't think you'll need like genius bar at the voting machines essentially.

Speaker 2 46:49

No, I don't think so. Well, actually, when we, when we did, uh, going out in the wild when Wisconsin used Prime three with, on the, on a touch screen, I'll never forget. I, I went to watch when they were using it, and I'll never forget I was in this, community center. It was, one of the precincts and this older guy come in, he's a farmer, and he said, I don't wanna use that thing. I don't use a computer. And then he, I heard him say, that's it. He had voted and it was the, that's, that was the response that meant the most to me outta everything.

Speaker 11 47:23

I'm really surprised. Thank you.

Speaker 12 47:27

I had a clarifying question. Mm-hmm. So, uh, with the VO platform, when the screen reader was saying, this printed out, was it based on the input or was it reading? What was visually coming out and interpreting that and telling it back to the user?

Speaker 2 47:45

It is not reading the visual image that is in the camera view. It's not doing that, what it's doing. So it

Speaker 12 47:54

still needs that, that text verification that you talked about later.

Speaker 2 47:57

Right.

Speaker 12 47:57

Okay.

Speaker 2 47:57

That's why we need that.

Speaker 12 47:59

Okay.

Speaker 2 47:59

In the event of, of confirmation there, I don't, the screen reader doesn't do well in trying to read that image. That's why I didn't even try it.

Speaker 12 48:11

Okay. It's like you didn't think, use a model. Habit, interpret the, and shoot it back to the screen reader.

Speaker 2 48:18

I, I could do that, but that was too much work.

Speaker 12 48:24

Fair enough.

Speaker 2 48:24

It, it was just easier. It's just easier to, to have the, the, the text up there and then they can hit enter and if they're blind, they can't see it. But then we had the second verification that came to their phone.

Speaker 12 48:39

Okay.

Speaker 2 48:39

And they were happy with that.

Speaker 12 48:41

Cool. Thanks.

Speaker 2 48:42

Yeah, I was, that was one of those, God, I don't have to do that.

Speaker 13 48:45

I, thanks. I, um, looks like Florida has come a long way from the days of the hanging, Chad, so kudos to you. But I was just curious, your comment about you haven't been able to sell it. Like what is, how do you try to sell it as a subscription or a price? Like what is the, what are the obstacles you're facing that more people, more states aren't interested in this?

Speaker 2 49:11

Uh, to be, I'm just gonna be bluntly honest again, it's about until you feel pain, we don't have to change. They

Speaker 13 49:21

just, but hanging Chad seemed painful

Speaker 2 49:23

and they changed that we had ballot marking devices, and even with the Michigan study, it wasn't enough pain to do anything different. I don't know if, in order to, to be honest with you, the, the, the way I think this could change. If an administration said, we're gonna put up money to have innovation in elections, and that's what happened with the ballot marking devices, United States Election Assistance Commission. you see that thing, the RAAV, there was a research alliance for accessible voting was created and we were able to work with vendors. So it would take maybe a national initiative or. It's just hard to say. They, they just won't move unless there's a state out there that wants to be on the cutting edge and try and do things. But even states don't want to, that's why they have vendors. They wanna go to the vendor and have the vendor, vendor manage that part of it. We'll see, I, I'm, I'm optimistic'cause you know, again, ESNS Dominion, they know who we are and what we're doing. we'll see. Okay. I wish I had a better answer. It's just, sorry.

Speaker 14 50:37

I just wanted to go back to the older adult, uh, question because 85 and older is the fastest growing population. A hundred and older is the second fastest. So I think it would be really valuable to really test in the oldest of the old. And secondly,'cause you say with touch screens. Pretty easy. And you're right, a lot of older adults know how to use touchscreens. However, as you age, your hands get drier. And so using a touchscreen becomes more challenging. And I've certainly seen it at clear through the airports of when they used to have touchscreens of them failing because the hands were dry and they would have lotion and creams, et cetera. So I just wondered if you were thinking about how to, how to modify for that and as well as if they have tremors.

Speaker 2 51:24

Yeah. even in, in our studies, we did see, with older adults there were some tremors and, and things with the tremors. Tremors actually were, I hate to say this, but helpful because they hit the screen harder. That was interesting. And the way we designed it, you can't drag and, and select a whole bunch of stuff. So we have some features in the interface to help against that. Uh. It's different from a phone. I think the, the phones kind of enculturated that population to being able to manipulate a touchscreen. But this big touchscreen that in front of them is different. They, they don't interact with something like that on a, a normal basis that you can't really tell. But that, that screen, uh, it's like 27 inches maybe. So I don't know what the perfect size would be. We'd have, you're right. I think we'd have to study that a little more. And I'm in Florida, we have plenty of older adults. we just haven't gotten, I don't have funding to, to do that study yet. if I get funding, I, I would definitely like to do that. I mean, I go down to the villages and hang out there and, and, and do a study. Uh, but I, I, I think based on the observations that I've seen so far. We haven't seen problems with this interface. The, the, the colors of the screen, there's no issues with colors. Colorblindness, double tapping hasn't been an issue like tremors. it's big enough and we could actually, I didn't show up, but I can make it screen even bigger, the ability to zoom. So we just haven't seen that. And maybe it's just, you know, knock on wood, we, we've been fortunate in that, but it's, it's, it's worked.

Speaker 14 53:09

Thank you. Kudos for what you're doing.

Speaker 2 53:11

Thank you.

Speaker 15 53:13

Hi. my question, one of the big issues in voting obviously is voter turnout. And I was wondering if you had maybe done like a post, if you asked people after if this would make them more likely to vote.'cause we have, you know, here in the US we do have a big voter turnout issue. And so if your technology would make people more likely to wanna vote again or become like returning voters,

Speaker 2 53:35

I think. The, the transparent voting machine, stuff like that, I don't think that makes a difference for, for, for that. But I think, the voter pass would, meaning the ability for me to mark my ballot on my phone, generate a QR code, and walk up to the machine, scan it, and have it print that that thing is engaging. I, I think that would bring more people.

Speaker 5 54:02

Thank you. One last question, please.

Speaker 16 54:04

Oh, okay. I have two questions, right? Perfect. Thank you. Thank you. At a setting choice, I'm really intrigued by this voter pass idea and the questions about older adults. I'm wondering about younger adults, youth, and I love the simplicity and accessibility and the, you know, the, the having no bells and whistles. I also wonder a curiosity about. Sort of the psychology, sociology of younger people interacting with government, really, you know, only a few times in their life and maybe only via voting. And if there's already sort of a lack of faith in institutions or a feeling that those are old fashioned or not serving their needs. I just wonder about the resonance of an object that feels really old to them. And I wonder if that came up in any of your testing, like if they were having an opposite experience of not being able to resonate because it felt. Clunky is not the right word, but old fashioned. And then my second question is, um, related to your last point there about, potential facial id. I wonder where sort of other countries around the world, Somali land, Ghana, Kenya, et cetera, are using IRA scans or, fingerprints. Like why isn't that a feasible option here in the us? Thank you.

Speaker 2 55:19

I'm so glad she asked that, so, oh, alright. I forgot to comment on that. to answer your first question, it's simple. No. Because voting only happens so infrequently and they don't even remember how they voted the previous time. as long as they could generate their ballot. And again, it's like, duh. And, and, and the only comments I got is like, this is for older people, but they did it as far the iris and fingerprint. the, those would potentially work technically. But again, from a psychological sociological perspective, they are not a possibility. Reason being fingerprint. What are fingerprints associated with criminals in the United States? That is a negative connotation. Iris. They don't, people don't wanna give Iris'cause they feel it's permanent. I can never change my iris or, you, you got me permanently. It's a psychological, I'm serious. We, we actually tested it. This is psychological and working with Damon on it. Iris and fingerprints are the biggest hurdle because of the negative association with fingerprints and Iris. The permanency of it, the thought of it. Facial recognition seems to be more acceptable because your face is exposed. I don't hide my face for the most part. My iris, they, I, I'm not hiding it. But technically you can't get to my iris is what I'm not saying. This is true. I'm just giving you the, the perception of why these things won't won't happen. So a lot of these things won't happen. It's not because of technical issues, it is these other things.