DayCast Transcript Episode 3: Discussion with Vince Daukas, a 17 year practitioner of AI solutions
Updated: May 24, 2022
Announcer: Welcome to DayCast, where we discuss how AI is really used in business, no hype, discussions with Real AI Users. Here's your host Corey Dayhuff
Corey: Today we have Vince Daukas with us here from the Dayhuff Group. Vince has spent the last seventeen years in cognitive solution design for IBM. He has been helping companies through their AI journey, defining what AI means for their business of building out AI solutions to meet their business needs. Vince has a unique perspective on hundreds of companies across many industries, welcome Vince.
Vince: Hi Corey. Thank you for having me.
Corey: Absolutely. So, Vince let’s start with how did you end up in AI? You’ve been in AI, probably as long as anybody I know. How did you get involved in that and what's that journey for you?
Vince: well, you know back in the early 2000s, we had the startup with a lot of different types of change in the IT world and in the world of understanding consumers and how we work with data. It was the age where the Internet was really cooking up a lot of ways to get data. You had big data. You had huge amounts of it, interesting forms of it, hard to understand collections, and people were massing all this stuff didn't know what to do with it. So, it became a big demand to try to understand that data better, and they started digging into some of the already established statistical and AI technologies that hadn’t been brought forward because the understanding wasn't there yet, on how to use it and starting to apply that, along with some of the great increases in computing art that we had to time, the ability to store everything, the ability to process things in new ways. That all started gelling into a better way to recognize patterns in the data and provide practical value. That was a little bit laborious and difficult in the early 2000s, but it started you know, leveraging things that helped us identify spam and do a better job of recognizing images and optical character recognition, and then it started branching into the areas of research where it can impact medical science better with understanding. And after that we started understanding how to deal with text better enough to where we could do something like compete on Jeopardy. And so by about 2010 or 2011 they created what was a convincing contestant on Jeopardy and that's what really kicked it all off. I think that opened the eyes and it did a great job of really showing how you could use this dream technology to actually provide real solutions and indeed do things that were like human behavior and even better than what humans could do so that was, I think, the kick off of it. And I was involved in a lot of the areas of IBM that were circling around those solutions, and that's how I got started in it.
Corey: So, was the acceleration of AI over the last really five to ten years, has that been mostly because of access to computing power?
Vince: It's been computing power. It's been new techniques as well. So, you know we didn't have the way to work with algorithms that were incredibly complex. So, the algorithms have been known for a long time and how to just recognize patterns and deal with dated certain ways, but they couldn't you know it would take days or weeks to process simple algorithms with old technology. The new technology could do it in microseconds. So that's what made it more practical. We've always known, or at least for probably forty years, fifty years we could add some incredible techniques with algorithms that could process data, we just needed something that could really do it quickly and make it so that we could make these things as complex as necessary to solve the business problem that was complex. So, the technology went hand in hand with understanding how to work with algorithms a little better, how to configure them so that they're easier to use easier to code and have more understanding based on statistical algorithms that were just broad math science and now bringing them into an easy-to-use user experience. So, it's a lot of different things that came together. We had to also have data sources and all of that kind of converge into making this practical for the real world and not just a lab experiment.
Corey: That makes a ton of sense. You've really been in and talked to hundreds of companies and talked to them about their AI journey and how they can leverage AI. What do you see is kind of a starting path or is very a common path that those companies use to start?
Vince: You know there really is some commonality in all this. Basically, AI is still a poorly understood area technology and business solution, and it is a range of understandings about it that you probably need to corral it to something practical. So, people see it as space age things that really are impractical and other people see it as something that they don't believe can actually do anything for them. I think the starting point is to bring AI into practical business sense, point out where it can actually be used to help a business do something valuable, and then, you know, start to pave the way to identifying opportunities in the company that would be the first things you'd want to tackle and some practical applications and give the understanding of how you actually bring this to life, to serve the need of a company. I think that's really the starting point for almost every organization, and then it builds from there.
Corey: Okay, so my guess is you identify multiple ways it can use AI technology and then try to pare that down to the best way. How do most organizations go about that, identifying here's the best way to start?
Vince: Well, this is an area that needs experts and there's expertise needed in a couple different areas. You need people that understand what AI is about, how it can be used, and you need expertise and how to envision a business or re-envision a business in terms of the transformative capabilities that AI could bring. So, a lot of people want to take their same business approach and try to apply AI to that, and you may be able to do some things with efficiency etc., but the real value comes from being able to reimagine your business in light of the new capabilities that AI could bring. And so, this is kind of a business expertise meeting technical expertise that's needed to be brought to bear in a discovery process.
Corey: What is the most interesting use case that you've seen a company actually implement?
Vince: Well, I’ve seen a whole bunch. And it depends on the industry you're talking about sometimes. I love the industries that help humanity directly with medical sciences and those have been some of the early areas of big lands and some of those are pretty amazing technologies to help say, drug discovery. I've seen where they've been looking at data and just doing investigations of the data to figure out the patterns that they see across research studies and this relates to today. You know you could say that some of the new development related to Covid, for instance, I’m quite certain that some of the r & a-oriented research is evolved from AI based capabilities that was able to take the look at thousands if not tens of thousands of studies and derive insights and patterns of those that allow the world to start to understand this very complex medical science. So, there's a lot of things and drug identification that I've seen happen that's just amazing, where, we say you know, we saw this classification of things work in another drug, and now, in this area we're seeing a gap that should be there because of the data pattern. And just by imagining that there's a gap, and there might be something there to explore they’ve discovered new drugs. And this kind of thinking way outside of the box is very exciting. Also, medical scanning, looking at images that have potential to discover cancer and being better than humans are at discovering cancer within medical scans, for instance, that’s just amazing that now it could do a much better job than a human can at consistently identifying masses or images that look that way. I've seen some incredible stuff in entertainment, where there, you know, one of the amazing things is that we create realistic looking people, realistic looking audio, realistic looking text that looks like a human created it, or looks like it is human like, and yet it's not, it’s created by a computer, and that's just astounding. We're going to see much more that going forward, where it is used to help increase the user experience by creating either from the familiar or interesting or effective, either visuals or audio or text to fool people, so that they can live their lives better. There's a lot of good examples of that out there. I've seen a couple of you know examples where there are companies that are using AI to do things completely differently. We had one company that was identifying gun shots in a city and triangulating those sounds by having microphones around the city and be able to tell quickly where there's a gun shot, having police move there, but they were getting confused with a bunch of back firings of cars etc. We took a completely different approach by creating visual imagery of those gun shots, or those sounds, and then analyzing with a visual analyzer so then an audio analyzer, and they came up with much better results. So, we're seeing AI start to take and do things that that are way outside the box using different senses that humans use to solve problems. So, this is now becoming outside of what we thought AI could actually do to really push it forward in to new realms that are not just simulating human behavior, but doing things well beyond what humans could even do. So, yea, I think there's so many different examples of you know, I think the single most interesting way to be today is they're producing brain linkages to movement and understanding how your brain works, it’s that a computer can tell you what you're thinking or induce action based on what you have thought. And so those are some, that's incredible going forward, and being able to have an understand of what you're thinking even better than you understand your own thinking is, I think, the future for this stuff.
Corey: Well, I can tell you whether it be a computer or anybody else, no one wants to know what I'm thinking most of the time. I don't even want to know what I’m thinking most of the time.
Vince: And I think the you know the image is here that it's going to have its own mind and use it nefariously. But really, we're talking about this AI technology is leading down a practical road. So, we want to empower you know the value factors or what’s driving it. So, where can we use technology that understands how you think, to give you as a consumer better information, to give you as a handicapped person better access to things that you could do or get to before? You know those kinds of applications are where we’re driving this AI technology, and it's not just off running on its own, to do things and in ways that aren’t good for mankind. So, I think that you know the process of finding valuable applications in this stuff in a business sense is really pushing AI into productive paths, instead of what the science fiction would make you fear.
Corey: Yea, I think that's the thing I've seen the most over the last three years is really using AI to make us all a little bit better, a little bit faster, a little bit smarter. You know it's less about the big wow stuff you can do with AI robots and all that stuff. It's more about how do I make the common day things that I do, how do I make myself better, faster and more efficient at that?
Vince: Right, and I think that a good example of that is in how we're now, you know I haven't pulled out a map in I don't know how long in my car. And you know, you're used to now hearing talking or hearing instructions on how to go places as you drive, and that, I think, is a really good example of how the future is moving forward. You'll see you know your hand-held device or even you know it might be a wrist device, something in your ear, that now becomes your concierge to life, if you like. And we're seeing that is the evolution of, say, chatbot technology, where it conversationalizes whatever interaction you're trying to accomplish so you can basically now start to talk to a device. You know, of course you have the Siris and all of those, but it goes beyond the mundane. Now we're talking about it, anticipating what you want, giving you advice, giving you recommendations, doing it in such a way that it's conversational, instead of laborious to get to, so we're seeing a lot of that use now of guiding people and helping people with conversational methods more and more and it’s accelerating incredibly quickly.
Corey: Yeah. I agree with that as well as kind of combining multiple AI technologies together to build something even bigger. You know. So, you talked about the IOT with the wristwatch activity. You know that connected with the speech recognition activity, that connected with the machine learning modeling around customer segmentation to identify who you are and present you with the right information. You have three or four or five different AI technologies, all together there, kind of combine to build a solution, that's bigger than one piece.
Vince: Absolutely yeah, we've seen quite a bit now compound technologies, but also seeing aggregations of capabilities across the Internet that are starting to be combined. So, you know you don't have to have computing technology yourself. You can just go out and have a process happen in the cloud, and you could have many different processes happen in the cloud. We're seeing businesses start up with business models that just take a process from one area in the process from another area and put them all together and create an incredible value that hadn’t been offered to the market before. We're seeing now the startup of aggregations of intelligence that's created for industries, and where you can't create perhaps enough intelligence in one big chatbot or one big AI analytical tool, and you can borrow it from other AI sources and start to have exponential increases in your intelligence, about how you're doing something. So, these are, you know, this networking and combining AI capabilities across clouds is now starting to create enormous, powerful AI capabilities.
Corey: Sure. What do you see is kind of the biggest return on investment, quick hits around AI for organizations that are maybe just starting down that journey?
Vince: So, the quick hits are right now, probably user experience based and research, so either experience has become one of those key areas where you want to be able to understand in real time exactly what a user would like to do and how they you know, what you could provide for them, that's going to give them the greatest satisfaction. And do it 24/7, 365, do it with real big consistency, do it with accuracy and build, you know the loyalty and the interaction integrity at that really will create and expand your user base. So that's what we're seeing as some of the biggest hits. So, wherever you’re touching users now using AI, especially in chatbot technology, that's really big right now to improve your way to deliver information to them. No more, you know the old model of having to dig through menus on websites or use, fill in big forms, or do things like that are laborious. that are friction points for user interaction. Now you want to start to have a coach, that's using conversational methods to help people so that they feel like they're just interacting in a normal sense, as if it was another person. Also, you know, ways to get to your data. Most companies are building big data sources and that's a great idea to collect all the stuff. But if you can't get to it, recognize what's important in it and then have it pull back exactly what's needed at the right time to the right person, then it's not very useful. And we've got now AI that can actually do that and then enable to feed either researchers, employees or users with exactly the information they need. No longer do they need to read pages of documents that you pulled back. You can set a right in on exactly the phrase or the specific information that meets your need. So, that's a really big important need going forward, that’s a quick hit for a lot of companies. And then also there's just ways to look at your business processes and automate those with better techniques using AI and increase your automation across the enterprise. I think that's a real big need right now. There are too many manual processes, there's too many things that people have to do that are mundane and routine. We can use AI now to recognize either screens or characters or text and have it make decisions without the need for very routine tasks to be bottling up or bottle necking your organization. So, I think that's those are kind of three areas I would say are initial starting points that are a big value for organizations.
Corey: So, in kind of going through that AI Journey, do you see a common number of mistakes or a set of mistakes that companies sometimes make even though they warn them?
Vince: There are actually. You know, it’s still new. AI has benefited from a lot of publicity to build the awareness of it, but it's also suffered a bit from being either, its being misunderstood either because it sounds too fantastic and people think it's you know are suspicious of it. Or they're not really understanding what it can do, and what its applications are, and that sort of thing, so that lack of understanding is actually kind of inhibiting things a little bit and it helps a lot to have experts come in and point out that this is a need, something that’s not outrageous, not something to fear, that's something that is very practical, that is very useful, highly valuable and then what it's about, how it works roughly and it's not doesn’t have to be mysterious. It seems like it's black box and we can do things now to help people understand even non-technical people what's going on here, so that they can start to embrace it. And I think that some of the starting points is to deliver that sort of understanding. And then the second part of that is there is you know, you need to help map how you would take those applications into the real world of their business and how that can be transformed to meet, how to get more understanding how to do that transformation of an enterprise for the future that would leverage all this incredible technology. Because I gotta tell you, you know the going slowly into new technology is not the right approach, because it's now that the potential to have competitors leapfrog and do things in such different ways that it's going to be disruptive in almost every industry. So, you know moving too slowly, not understanding, not embracing practical methods, are all inhibiting factors. I think that we can help and there's lots of expertise out there to leverage in getting you down that road.
Corey: Do you think lack of understanding is really kind of the biggest inhibitor for most companies in implementing?
Vince: I think so. I think that's the biggest inhibitor because once you realize what it is about and then what it can do, it's a pretty easy task then to figure out how to start making changes that can help your business. That's a big stumbling point and you know just envisioning that it's too big to tackle. I see that a lot, it’s like wow, this is all wild stuff and there's no way we could do it as a small business. Well, that's not really true. Actually, this very quick start easy to use, tasks, ways to get going. And then the other extreme is don't believe, I believe I can do everything with this stuff, and can it just take my company and run it, you know from all aspects? Well, that's kind of too big bang. It doesn't make as much sense to you know, sure there’s a lot it can do, but it's not going to do sometimes what everybody thinks it can do and there are some very specific ways to start with this, and corral your thoughts into something that’s got the right starting point and build to, you know, more and more value. Those kinds of thoughts are inhibitors because they don't really let it get off the ground in a way that works for companies.
Corey: You know AI from an implementation perspective, is almost perfect for agile methodology, because you can get quick hits and you can build on those quick hits again and again and again. And I think what we see sometimes especially in large organizations because of their historic way of doing funding for projects it's almost that they have to put together a large initiative before they can get any buy in to do anything. And I think specifically, if they take, they take a small, tiny project, a skunkworks, looking kind of project to management, they don't get much traction because it gets lost in all the other large projects that may be going on. So, it's almost a, they almost hurt themselves, in not being able to do maybe some more advanced things that drive real ROI that are inexpensive. They're only looking at the big things that they're doing, not some of the small things they're doing that may actually alleviate some of that big stuff.
Vince: Right. That's very true that lots of times there's a better way to think about all of the big re-engineering projects you have based on some of this technology. But also, you know, there's the problem that it looks like too much like a lab experiment. It doesn't get management attention as well and you want to bring it to life, you know and be able to show management here's some real, quick and easy ways in which I can show you this works. And there is an art to that. And so, you want to bring this the business users in a way that just doesn't sound like it’s a shiny new object in scientific terms. It's actually got practical business sense about it.
Corey: Absolutely. Hey, Vince I want to thank you for doing this today, very enlightening. I'm sure we could go on for another hour with all the use cases and talk about specific companies that you've dealt with in the past. But once again thank you so much for being on DayCast today,
Vince: My pleasure Corey. Thank you very much for having me.
Corey: I love it, thanks again.