#181 Bringing state of the art conversational AI to Market with Dr. Jason Mars, Associate Professor Of Computer Science, University of Michigan

---

In this episode, Felipe is joined by Jason Mars, professor of Computer Science and Engineering at the University of Michigan. Also, an entrepreneur and an author, he cares deeply about creating an impact in the world. Describing himself as a tinkerer at heart, he has always tried to experiment and explore, with an ultimate focus on making a difference in the world. His background is in artificial intelligence at scale, having worked that on a large-scale computing system for many years. He has written about 100 research papers that worked with 14+ PhD students. Having started several companies, he has grown one technology company to a $200 million valuation bringing state of the art conversational AI to market. In this podcast, Felipe and Jason discuss creating the kind of technology that can define the next 10 years of artificial intelligence and scalable systems.

Thanks to our sponsor Talent Insights Group!

“I’m a tinkerer at heart. And I’ve always tried to experiment and explore and let my curiosity drive me in different directions, but ultimately focused on making a difference in the world and moving our species forward.”

— Dr Jason Mars

Episode Highlights

We've been on this crazy journey just for the last 100 years or so, of really engaging computing technology and, and understanding after inventing, understanding the internet and understanding how we can always be connected, and it's part of our, our deepest, most psychological, you know, desire to be able to engage with technology in the most natural way possible. It turns out that we engage with each other as intelligent beings, engaging with intelligent beings, we have conversations with each other, we grow from those conversations, we get inspired by each other.

I call 2010, the year of the AI revolution. That's the year that Google did ImageNet. And they did a project this was a deep learning project is one of the first times that computer vision really took a deep learning focus approach to solve some holy grail image problems.

In the NLP space, conversational AI space, the takeover of deep learning was a little slower, right, it took a few more years for deep learning to dominate the language, conversational AI space and vision, and fundamentally others would argue with me, I believe that the conversational realm is actually very hard.

I mean, I'm fortunate, lucky that I'm able to see this technology space from two angles. All right, there's a perspective that comes from understanding the market and the kind of products that folks are engaging in the market and what is having success, and then also look at the science, from a research innovation perspective as to what's being published and publishing papers that are innovation myself. So from those two perspectives, it's actually a very interesting landscape, right? Because the kind of technologies that people are engaging every day with conversational AI does not represent the possibilities that come from the scientific discoveries that are going on. Now, this gap is closing. And like, it's crazy, because it's right, right before my eyes, like five years ago, the gap was larger.

A Transformer is a type of, sounds cool because it's like Transformers, robots in disguise. But it's a type of neural network model in the academic realm of spirits, a kind of neural network model, that the reason why it became so interesting is that it was clear that recurrent neural networks are important in this space, which means the kind of neural networks that looks at the order of things, the sequence of things because language is a sequence, it matters the order the words are in a sentence to understand what it means.

I would submit, my conjecture is that the tech is there, if people invested energy to create experiences that are where the users are. Right. And so, you know I think we're gonna see more of that coming real soon.

So we're gonna need that kind of disruption. It's risky, but we're gonna have to wait for it, you know, which is crazy. But what we could do is encourage people to think disruptively, we can encourage young folks to take risks. We can create a culture where we let people make mistakes, and we teach them how to have grit. Grit is extinct as a concept. Now in today's world, it's so sad, but we have to teach grit, because that's the mill you that creates the right kinds of innovation, to get us to break out of this stagnation never end you know?

I mean, that's a very deep thing because people do get excited about tech. But they forget that to create value in this world, you've got to fill a void of pain, or an opportunity to delight humans in a sustainable way.

Tell me the story, the narrative of the void that you're filling for some human that is suffering? Or how are you bringing delight to some human?

What I find fascinating about your focus is that you're on the cutting edge of innovation in so many different realms. So the focus on removing pain, emotional pain, or delighting a human, that's, that's something that is happening. I guess, kind of like, slowly, it's a trend that I'm seeing slowly growing business. One, but definitely not happening in AI.

And this is advice to all the folks over there out in the world. Like you, you should, if you have a spirit that you're always pursuing insights, you're always looking for phenomenal mentors, that that you can learn something from, that's where you build these things on the journey. Somewhere along the line, this insight, we became very clear and, and integrating, integrating and asking the questions as to what makes A successful and what makes B not successful, we'll let you get on that journey of discovering how to be better, think better, and to change your mental frame or change the model you have for the world.

But life is this journey of doing this modelling, right and learning from our experiences and learning from our mistakes. And you have to get out of your comfort zone and make mistakes and get challenged, you have to be challenged in order to improve that model. And so that's where it comes from. It comes from a journey where I always try to test ideas, I test my own dogmas and certainties.

And then you discover these nuggets that are gems that help you make good predictions. And that's actually how I run myself. And that's I attribute a lot of my success to that. And you can call it learning quickly or learning how to learn or learning how to grow. It's to that and it's to grit. That means you persevere. You mix these two things together, perseverance and an ability to adapt and learn and you can achieve any of anything you dream to achieve in life. I'm certain those are the two elements, grit, learning how to learn openness, conscientiousness

At Data Futurology, we are always working to bring you use cases, new approaches and everything related to the most relevant topics in data science to help you get the most value out of these technologies! Check out our upcoming events for more amazing content.