UB Insider #45: In the Midst of the AI Revolution
About this episode:
To some, artificial intelligence sounds like a science fiction concept that humanity is perhaps on the cusp of discovering…but chances are you’ve used at least one form of AI today. According to a recent study from InsideSales.com, most people aren’t aware they run into AI every day. In this episode of UB Insider, Judd Bagley, director of corporate communications for InsideSales.com, tells about the study, how demographics come into play, and what the results suggest about the future of AI. You can find the Wired article Bagley mentions here. Subscribe to our podcast or download this episode on iTunes, Stitcher or Google Play.
Lisa Christensen: Hello and welcome to UB Insider. I’m Lisa Christensen, Online Editor at Utah Business magazine. Artificial intelligence is being utilized in a wide-range of fields from medicine to customer service, but a recent study from InsideSales.com suggests that most people think artificial intelligence or AI is still a science fiction concept only just starting to become a reality and are unaware that they’re already using it. Judd Bagley, Director of Corporate Communications with InsideSales.com is here to tell us more about that study and what it means for the workplace now and in the future. Welcome.
Judd Bagley: Thank you very much for having me.
Lisa Christensen: So Judd, tell me a little bit about this study. Why did you do it? What insights did you see?
Judd Bagley: What really sets Inside Sales apart as a company is our capacity to make software that makes smart decisions based on a tremendous data set and smart algorithms that come together to help our clients do a better job basically accelerating the sales process using artificial intelligence.
Now one of the things we find is there’s a wide range of, a big spectrum of understanding of what AI is that ranges from people that are avidly trying to incorporate it into their lives and other people that are afraid of it. We want to understand a little better who these people are, how they view it, who they are demographically for example. So that would basically give us a better sense of understanding of how to sell and represent our product.
We commissioned a survey of about 2,200 people in the United States and the U.K. and asked them a series of questions. One of the most interesting findings to come of this, although it probably shouldn’t come as a huge surprise, we started out by asking people how often they use AI and then went through a series of other questions. And then we ended by asking them, how often do you use these specific applications in your life? And what we found is that there’s a great disconnect between the amount of AI that people think they use and the amount that they actually do use. So that may mean that AI is doing its job. In other words, it’s integrating into people’s lives seamlessly in ways that they’re not even aware of. And so based on that kind of foundation, we then looked at some of the other data, cut it up, cross tabbed it and figured out a lot more about how people view AI, how it’s affecting their lives and what they see the future of it being.
Lisa Christensen: One of the interesting aspects of the study that I hadn’t thought about before or wouldn’t have guessed is that according to the study, 78% of people in higher income brackets – $175,000 to $199,000 – said that they hadn’t used AI while a quarter of those who said they used it all the time were in a very low income bracket, under $27,000. Why do you think that is, and what does that say about workplace demographics and their cognizance of AI?
Judd Bagley: One of the things we really wanted to understand was the generational breakdown. So we went out of our way to sample Gen Z as they call it, then the millennials and the Gen Xers and then the baby boomers to see how everybody views it. This is due in large part to the fact that millennials very recently became the majority cohort in the workforce. And in the world of sales, they represent the vast majority of people in the workforce. These are kids who, I say kids, who’ve just come out of college and their first job is software sales, for example. They’re trying to work their way up in that industry and they’re starting in sales.
We’ve found that accommodating the millennial mind is absolutely vital to the success of our product. So we wanted to better understand how millennials see this and then break that down across the other generations as well. What we found is that, and we probably could have predicted this, Gen Zs and millennials are much more avid users of this technology than the Gen Xers and the baby boomers. They’re also much more aware of the fact that they’re using it than Gen Xers and baby boomers.
So when we looked at that income distribution data and saw that it’s basically U-shaped, that the highest use of AI comes from the lowest income, like $10,000 and under, and the highest income as you referred to. What we hypothesize is that this represents kids in college who are working their part-time minimum wage jobs, who live on their phones and understand what they’re using. That’s the low income. The high income represents people who are probably high up in the technology field that are higher income and they’re more comfortable with it. They also understand that they’re using it. What was interesting was that the income bracket that represented the lowest use of artificial intelligence was $125,000 – $150,000 a year. Which is still pretty decent, yet we figured they fall kind of right in the middle of those two demographics. So as far as we know, that’s probably what accounts for that odd distribution.
Lisa Christensen: So your study also asked and then noted about everyday uses of AI, you know, you’re saying how often do you use these applications, and obviously these things are all around us. Where are some of these very common applications that many of us might use without realizing that we’re interacting with artificial intelligence.
Judd Bagley: So when you look at where the greatest use of AI is, it’s sitting in your pocket right now. So those mobile-based applications, those that are in your phone represent by far the most AI use. So those are specifically traffic navigation applications like Google Maps and Ways, it’s streaming music recommendation services like Pandora for example. It’s Siri also. So those kind of artificial, or rather those digital assistants as they’re being called of Siri, Google Home, what’s the other one? Amazon Echo. Those are by far have the most use and those are the things that people kind of carry around with them all the time. And then it falls way off after that and you get into applications like FitBit for example, or home automation software and things like that. So the telephone ends up being, the mobile phone ends up being the real kind of vital platform for this kind of technology.
Lisa Christensen: Why do you think, given how common it is and how much we use it every day, do you think people are so uneasy at the thought of using artificial intelligence?
Judd Bagley: Well when you get down and look at the data you see that they are actually quite at ease with using AI for you could say low-consequence activities. So entertainment recommendations for example. They’re overjoyed to have Netflix suggest a movie, and if Netflix gets it wrong that’s two hours of their life they’ve lost. Big deal. Where their level of dis-ease jumps way up is when it comes to much higher consequence activities such as for example, medical diagnoses. Which interestingly would appear to be one area where this technology is really going to make a huge difference.
People aren’t very trusting of that. They’re also not very trusting of things like hiring, automated hiring, financial advising, the idea that artificial intelligence might move your money around on its own in ways that it views as most beneficial to you with regards to the market. Also, autonomous cars. People are not yet comfortable with that. So in all four of those areas, the consequence of something going wrong is quite high. When it comes to entertainment, even an application trying to tell you what the best way to get to work is, even if it gets it wrong, that’s 20 minutes of your life you’re out. Big deal. But the things like medical diagnoses, elder care, hiring, financial advising, people are not very trusting of that yet.
Lisa Christensen: So what impact do you think this unease on development of AI and implementation of AI in those areas?
Judd Bagley: Well it’s going to be a slow process. But as the data suggest, those emerging generations, the Gen Z and the millennials are much more eager to integrate this stuff into their lives. So this is an inevitability. There is a wave coming that software application developers are anticipating and are going to be developing to them. Nobody is creating new products for the baby boomers and less and less every day for Gen Xers. It’s the generations coming after them who are much more willing to take a chance on these things, to incorporate them into their lives. So it’s going to happen. Even if it were to happen today, older generations would probably not be the ones to integrate it into their lives, but the younger ones. So it’s going to happen.
Lisa Christensen: For these older generations, as AI does happen within their lifetime still, how do you think that sentiment of distrust can be overcome?
Judd Bagley: You know, that is a great question and marketers don’t actually spend a lot of money marketing to those generations because baby boomers for the most part at this point have probably settled in on what their favorite products are, what they’re comfortable with. And that’s not likely to change so much. I don’t know that that’s going to change, but time will tell.
Lisa Christensen: Looking at the future of AI, what does it realistically look like?
Judd Bagley: The future, where the real opportunity is, and this is actually a pretty interesting topic. You look at those applications that have very successfully incorporated AI so far, they are Facebook, Amazon for example, Google, Pandora, these are all B2C enterprises, businesses providing services to consumers. The reason why they have been able to get out in front of this so early and have had so much success in incorporating this is because for a B2C company, the potential market is gigantic.
So Facebook for example, their potential market can be numbered in the billions. The same is probably true for Amazon and the others. In order for artificial intelligence to really work, it’s not so much just the math or the algorithms. That provides the framework. What you need to do is pair the framework, the math with gobs and gobs of data that can then make the algorithm truly intelligent and provide truly useful information. That kind of data has only existed, that level of data has only existed in the B2C state.
When you look at people’s use of AI and what they report, it drops off, it drops off a cliff when you ask them how often they use AI at work. Only 20% say they regularly use AI at work. And we suspect, we probably could have designed this study a little better because we suspected what a lot of those people were likely referring to is Google. You know, they’re sitting at work and they use Google. As far as actual artificially intelligence workplace applications, there is just about nothing. Inside Sales is an exception and I’ll get to that in a second.
So what does the future look like? The future, as businesses acquire more and more data and finally hit that scale where it becomes really useful in a mathematical construct, you will begin to see many, many more artificially intelligent applications made available in the workplace. So that’s what the future looks like. That’s where things are really going to change enormously. Now let me address that for a second.
What makes Inside Sales unique and one of the reasons why we’ve had so much success in the business to business space is because we’ve basically convinced our users to pool their data. So you take our thousands of customers and they all have bits of data here and there regarding their interactions between their sales people and their potential customers. No one of them has reached that scale where it is significant and useful. But if you pool all of them together, suddenly you do hit that scale.
What Inside Sales does, we’ve amassed a trove of I believe we’re up to about 110 billion sales interactions that come from all of our clients, so all of our clients essentially pooling their data for mutual benefit. This is the kind of paradigm that is going to have to prevail in the short-term at least for artificial intelligence to find success in the workplace. This is going to happen more and more often as time goes on. So that’s really where you’re going to see the biggest shift. And that’s where we believe the greatest commercial opportunity is in workplace AI.
Lisa Christensen: And it sounds like that kind of collaboration is going to have to happen in all kinds of fields.
Judd Bagley: Absolutely. Yeah, if medical diagnoses really are going to become a thing where AI is concerned, the government is going to have to figure out a way to get this all in one place, to anonymize it, to preserve people’s privacy rights and still use it for the benefit of all. So yeah, that’s just an example. It gets thorny in places like that. And I think one of the reasons that we’ve had early success is because you don’t have those kinds of privacy implications. We take the data that’s offered by our clients, anonymize it, pool it, normalize it and then begin to derive useful conclusions from it. That’s also kind of a lower consequence activity than pooling people’s medical data.
Lisa Christensen: Yeah, what kind of policy changes are going to have to happen in order to be able to pool some of that more personal data whether it be medical or financial I would guess is similar, or any of these other higher risk applications.
Judd Bagley: You know, this is kind of interesting. Because when people refer to HIPAA, that’s become shorthand for just patient privacy. Which is really what it is, but a lot of people aren’t aware of how HIPAA actually came to be. Kind of during the Clinton era it was recognized that if you could pool all the treatment and outcome data for all kinds of illnesses, you could make better decisions. And so HIPAA was kind of originally seen as a way of pooling data while preserving patient anonymity in order to achieve better outcomes. Now I don’t think HIPAA anticipated artificial intelligence, right? This was people who were looking at it and doing their own sort of statistical analysis and figuring it out. So a little bit of the groundwork has been done there. There is precedent. In finance, I think that’s a very, very tricky space.
Lisa Christensen: Well I guess that’s something that will have to be developed before AI can be more widely developed.
Judd Bagley: Exactly. So if you want to learn more about this idea, there’s a writer for Wired magazine. He’s an editor named Cade Metz. He is doing some of the most compelling reporting on AI that’s being done out there. He really gets it. He did a story in January, he did a series of stories actually about the use of AI in poker. Now one of the things that’s important to understand is that logic trees, you know, if then statements, are not really new.
In 1959, there was an academic who created a program for playing chess. Pretty simple, right? And it was successful in its way. So this sort of thing has been done. Where it gets complicated though is when you throw in the human element. Poker is one of those games where not only do you have to understand the implications of your own cards, you have to understand the implications of the limited data, the limited information you have, the imperfect information you have about other players’ cards. But not just that. You have to anticipate how they will perceive the information they have and the information they have about you and how they will likely behave. That’s where poker gets so fascinating and interesting.
So Cade Metz wrote a series of articles about AI in this Texas Hold’em tournament. What has made the difference is that these researchers analyzed thousands and thousands of Texas Hold’em games, I believe that the addition of, the arrival of internet-based gambling probably makes this data collection much easier. They analyzed this data not just from the standpoint of the pure mathematics of I’m holding this card and this is what I know about this person, but from the standpoint of how is the person on the other side of the table, how is my competitor likely to behave?
So this is the big challenge is anticipating the behavior of humans. Math is not very good at that. However, if you can, there’s this principle of human nature, maybe psychology is what you’d call it, called propinquity which states that under similar circumstances, similar people will behave similarly. So if you can put together a series of profiles, that basically represent kind of certain types of people, and if you can put the person you’re working with, if you can assign them to a profile and you can use the data to help inform how you believe that profile will behave under certain circumstances. You can make decisions that go far beyond the kinds of decisions that humans can make. Even those with the best intuition.
So this is kind of what happened in this series of poker games. They added in that human element. They would assign the other players these profiles and then using the principles of propinquity would use that to predict how they would behave under certain circumstances. And when you threw that in, suddenly the computer became unbeatable. Just about unbeatable. Even these professionals would kind of say they would find holes in the game’s algorithm, but then by the next day those holes would be closed up. In other words, the computer would learn from the mistake it made. Similarly, this is what we’ve been able to accomplish at Inside Sales, is we’ve created millions of profiles for people on the other side of sales transactions.
So if you’re selling software, for example, from one business to another, if you can know something about the other side of the transaction: about them demographically, about the nature of the firm that they work for, about where they are geographically and about other outside data like, you know, like the price of gas and what traffic was like that morning, and whether it was raining in that location or not. You kind of put all these things together, and the software makes really smart recommendations based on what we know about that person on the other side, the circumstances in which they exist and how they are likely to behave under circumstances that are similar to when that same profile was observed to behave in many other circumstances. So that’s what makes our software unique, is we’re able to account for that human component.
Lisa Christensen: That’s fascinating. Well we will have a link to that story you mentioned as well as to the study on our website. Thank you so much for coming on.
Judd Bagley: Thank you very much.
Lisa Christensen: Thanks also to Mike Sasich for production help today. Tell us what you think at firstname.lastname@example.org or you can contact us on social media at @utahbusiness. You can also subscribe to this podcast or catch up on old episodes on iTunes, Stitcher, Google Play or wherever you get your podcasts. Thanks for listening.