The Q&AI: Computer Vision and Real-Time Analysis. The Power of AI.

Bob: Welcome to a special edition of BoB Friday Talks, coming to you from Las Vegas at Juniper's Executive Circle CIO event. And today we have a special treat of having Dr. Gina Guillaume Joseph join us today from Camio to really talk about the convergence of two great technologies, computer vision and large language models.

Well, welcome Gina. So maybe we can start the discussion with an intro on your background. How'd you get here?

Gina: Well, I got my undergrad in computer science from Boston College, my master's in information systems from the University of Maryland, Baltimore County, and I got a PhD from George Washington University and thus began my AI journey.

I built a predictive analytics model using logistic regression. To predict software project failure and what are the factors that cause projects to fail. And so, this was in 2016. So early on, almost 10 years ago.

Bob: Okay. Yeah. By then AI was kind of a thing already. You know, I always tell people that AI is in the eye of the beholder. Last month I was in Scottsdale. I got into my first Waymo, Uber, no driving. I asked you, is that AI? For me, it's like when that happens, like, this is like real AI. So maybe we can give the audience a little bit, your deficit in AI, when did it become a real thing for you?

Gina: It became a real thing for me while I was studying and getting my degree. But AI is the, study of machines and software and ensuring that they operate with it like a human. So, mimicking human behavior. It's a subset of, machine learning and, and so we're going to talk about computer vision. We're going to talk about deep learning and some of the things that Camio does. So that's, that's where my journey began.

Bob: You know, when I talk to people who know all about AI, it sounds like you, because you started your education, when did you first go to school back in the 2000s or something?

Gina: Yes.

Bob: I don't think AI was really a big thing back in the 2000s.

Gina: It wasn't.

Bob: We all started our careers with machine learning and regression and all that stuff. How do you describe the difference between machine learning and what we're doing now?

I mean, when do we go from just regression models into something we think of artificial intelligent and cognitive reasoning?

Gina: Yeah, so Alexa, Siri, those are things, Watson, the Jeopardy game, it'd be Jeopardy. So, I think we started there, but it's a progression. I think the earliest models was, you know, again, mimicking humans, having machines mimic humans in the fifties.

And then we progressed to where we are. And now we're in the ChatGPT explosion. We're looking at natural language, chat, generative, pre-trained transformers looking at, how we can, predict the next words in a sentence, tokenization and, things of that nature.

Bob: Yeah, it's funny you mentioned Watson because,that was one of the inspirations for me. When I saw Watson beat Jeopardy, I was like, hey guys, if they can play Jeopardy, I should be able to build something that I can play networking, right? I can do something on par with IT teams. Now maybe we get a little bit about Camio because I know you're the CTO of Camio and, you know, what we discussed earlier was.

It looks like Camio has brought together two of the earliest AI technologies. Computer vision probably was the first generation of AI doing amazing things, identifying things. And then we have these large language models.

Gina: Yeah. So, Camio was developed by an ex-Googler, Carter Maslin, and this was in 2011. And what he wanted to do, he wanted to create a video search similar to Google search engine. So being able to look at video and imaging and pull out in data and insights so that you can make decisions in real time. And so, what Camio does, it takes natural language, interprets it into an AI based rule and then allows the users to go in and leverage it for unauthorized access, for unsafe behavior, to be able to make real time decisions to notify the users that there's some deviation in the policy and, you can make quick and swift action.

Bob: Okay, so we're using computer vision here to identify objects, and we're using the larger language models to basically look up the policy, and we're trying to combine these two together, basically bringing some sort of computer vision together for large language models.

Gina: Yeah, absolutely. That's what Camio is doing. It is helping in, ensuring the safety and compliance, security in organizations. One of the use cases, that we're leveraging it for is, you know, we're in Vegas. So, unauthorized access in these buildings, in the hotels, it can identify unauthorized access. Flag and say, hey, someone entered.

Or, you know, they tailgated without a badge and go, mitigate that deviation and follow the user and ensure that they're not a bad or threat actor coming in. And one of the things that Camio did was in a large, hotel like setting here, someone fell into the pool and they were able to identify that person falling into the pool and immediately flag the user who then brought in security and were able to pull him out of the, him or her out of the pool, and save a life.

Bob: Okay, so when I go down to the gaming floor, am I going to have Camio keeping an eye on me, walking around, or where's Camio deployed right now.

Gina: I don't know if we have it here at the MGM Grand, but residential buildings, we have it deployed in universities, we have it deployed in different companies across the nation.

Bob: Yeah, so given your background, you've been working with computer vision, you've been working with large language models. Can you give the audience your vision of where you think we're headed? Computer vision, large language, what do you think is next on the horizon?

Gina: Oh, my goodness. AGI, you know, getting computers to be more human, like to have emotion and feeling, I think that's kind of where we're going.

Bob: Okay. So, this sounds like Star Trek type of thinking right now.

Gina: Yeah.

Bob: We're getting close to where the computer actually, I can talk to my computer and my starship, and actually get it to have an emotional response?

Gina: Not yet. But AI is moving so fast that, part of that is putting guardrails around it policies and governance and the federal government here in the US is doing that European Union is doing that because it is kind of like a runaway train and we don't know what we don't know. There's so many unknown unknowns and we're here at this Juniper Summit to talk about what are the implications? What's the future look like for AI?

Bob: Yeah, that's another discussion that's going on right now between regulations compliance around all of this computer vision, large language models between what's going on in Europe and what we're seeing in the US, which seems to be a much more distributed model.

Gina: Right?

Bob: You know, any words on what you see happening with the compliance?

Gina: Yeah. So, I worked for MITRE Corporation for 10 years. MITRE operates federally funded research and development centers, and I worked across the federal government, helping them in their technology transformation journey. And part of that was ensuring policies and governance around technology, and we know that the White House is trying to do the best that it can for the, you know, AI policy. They have the AI Bill of Rights to ensure fairness, transparency, data privacy, all the ileitis that ensure that AI is safe for humans.

Because, it's to make our lives easier. But there also can be some unintended consequences and being able to put the rules and regulations around that as it's moving so fast. Again, we don't want to stifle innovation, but we also want to put rules and guardrails in place to, to protect humans and to protect the planet.

Bob: Well, I have to say, in terms of the speed at which this is happening, to be honest, I didn't think I'd be in a self-driving Uber before I retired. Right. So, maybe you get any predictions on I always tell my friends, make sure you save that driver's license because it's going to be an antique.

So, at some point, we're going to prove that these autonomous vehicles are safer than humans and they're going to make it illegal to be driving. Any predictions of how far away we're from that?

Gina: Oh my goodness. I have a Tesla, so I have not turned on the autonomous feature. Because I'm human. I'm still fearful. And we are fearful. And that's part of what we have to manage, the fear of AI systems and them taking over the world, you watch the movie iRobot and just the, how they came together with reasoning and thinking and took over the world. But again, we, we have to, move with AI with, caution almost, right?

But not too cautious that we stifle innovation. We stifle the forward movement because it's going to move. It's going to go, we're going to be in a place in 10 years where we can't even fathom right now.

Bob: Yeah. I mean, that's why I tell people, ultimately, we're now building things or doing things on par with cognitive reading skills, and you almost have to treat these new AI assistants as interns' people. And I've had people tell me that, ChatGPT is the sarcastic parrot. What happens when it hallucinates? And I tell them like, what happens when your kids or employees do something that doesn't make sense? They're hallucinating too.

What the heck were you guys thinking when you did that? So, I think we are getting to that world now where we have to learn how to trust and understand how these AI systems work.

Gina: Yes.

Bob: So maybe a final word for our audience on. Where are we going to be in five years with Camio?

Gina: We will be able to use it in different use cases. So I'm talking with some folks about using it in the medical space, ER and the OR, to track doctors and the medical staff and performing surgeries and it allows the doctors to do their charting and reporting almost in real time as they're doing the surgery. But 10 years from now, I wish I knew. If I knew, I think I could play the lottery and win.

Bob: Well, it's funny you mentioned healthcare, because I was just talking to the CIO of Emory the other day, and that's exactly what he's talking about, is using AI to basically improve patient experience and offload doctors. Basically take all the work that they're doing right now and leveraging that.

So, Dr. Gina, I want to thank you for coming. I want to thank the Bob Friday Talk audience for joining us, and look forward to seeing you on the next episode.