Behavioral AI -How Good is AI When it Comes to Nonverbals?

Rachel Cossar: And welcome to yet another episode of conversations in the future of work where we welcome on specialists across a variety of different fields and spaces and areas to discuss their thoughts and their views on trends we’re seeing in our somewhat chaotic world of work these days In this episode, I am thrilled to welcome on Skyler Place who has a very specific area of expertise, which is, of course, near and dear to my heart being a nonverbal communication expert. So without further ado, Skyler Place. Over to you. Hey, Rachel. Pleasure. Thank you so much for taking the time to join us today. So just for the audience, give us a few, a a tight overview of of your your bio. Sure.

Skyler Place, Ph.D: So I’ve spent most of my career working at the Intersection of Psychology and knowledge. So how can we use tools like artificial intelligence to improve the human condition and to help people communicate better control their health better and interact with their peers in a more effective way. And I’ve built a couple different technology companies that really leverage the power of the voice. So what can we understand from how people communicate, from settings ranging from veteran mental health all the way to customer support in enterprise business. Awesome. I’m thrilled to have you on. Thank you.

Rachel Cossar: So jumping jumping right in, I’d love to hear a little bit more about because I mean, of course, you you you did your PhD in a very interesting space. And then the world has evolved in a very interesting way. So what are, what’s, like, most surprised you about the application of your PHE and Yeah.

Skyler Place, Ph.D: So so my undergrad was in computer science, and my PhD is in cognitive psychology really focused on quantitative modeling. So how can we you know, leverage what has become artificial intelligence to understand human psychology, human intentionality. And it was interesting. So I graduated roof almost 15 years ago, with my PhD. And in the, I would say the 1st 10 years, the focus and the skill set that was most relevant was really on the data science skills. So a lot of folks with quantitative PhDs were being, hired in the tech companies and leading data science efforts, leading quantitative research efforts. And what’s been really surprising for me in the last couple of years, really since the whole explosion of generative AI is when I’m talking to CEOs and advising organizations on how to think about their strategy, the philosophy of AI has become the focus point of many of their questions. So things like What are the unique skill sets that humans have that I don’t want AI to be replacing? How can I leverage the strength in creativity and critical thinking that humans outperform computers in and combine that with some of the generative AI technologies to coach people to be even better? So it’s been really interesting that, you know, my background standing philosophy of minds and sort of human intentionality more on the psychology and philosophy side. I’ve been leveraging a lot more recently. Then the kind of quantitative mathematics that was the foundation of a lot of the, modeling and machine learning work that I did earlier in my career. Yeah. That’s fascinating.

Rachel Cossar: It’s like now that the technology is really here knocking at our doors, these more, nebulous questions around, like, the interaction between AI and human minds right, is is at the forefront. Absolutely. Yeah.

Skyler Place, Ph.D: Because you’re seeing folks are are pausing, which I I love. They’re pausing and thinking and saying, I don’t just wanna replace my workforce. I have some amazing employees who have some unbelievable skill sets and experiences. How do I leverage that? How do I leverage our unique abilities as humans and and help people excel even more, with some additional tools? Totally. Absolutely.

Rachel Cossar: And, so on that note of how to leverage AI to amplify and and help people improve as as you mentioned. You know, what are your thoughts on, AI’s actual ability to provide meaningful feedback on like human behaviors. And and I say in real time. Doesn’t have to be in real time, but I know that that’s part of your your background as well. Yeah. Absolutely.

Skyler Place, Ph.D: So I’ve been working in the space for 15 plus years now, and there’s been solutions that can understand human behavior, from an AI perspective, if you will, for at least the last 15 years, and there’s been academic research going on for a long time before that. And the the real time component is is an interesting part because I think the the caveat with real time is that it is very computationally expensive. Yeah. So when you think about providing feedback in real time, it’s on average, you know, 10 times more expensive than, providing feedback with the delay. So you have to, as a a business leader really think about, you know, what are the use cases where real time gives an unbelievable advantage for your product and for your users. Because you’re gonna have to be charging a lot more in order to to have a profitability with a real time solution with today’s, you know, technology. Yeah. Go ahead.

Rachel Cossar: Well, I so is that mostly because of, like, the the server side processing costs and how quickly you kind of have to get things from one end one end process analyzed and then, like, regurgitated back to the user in real time. Exactly. Yeah.

Skyler Place, Ph.D: So it depends a a lot on your your stack, if you will, your your technical stack. So if you can do the processing at the end user device on a phone, for example, there’s ways to make it less expensive. But in a traditional big enterprise approach of you are streaming data from a data center to the cloud processing, sending it back down to end users, who often are on thin clients in corporate environments. They don’t they don’t have local, compute capabilities. Right. Then that’s where that’s where things get very, very expensive in in an enterprise setting. But what’s interesting is despite all of that, the the accuracy, the impact of these solutions on a whole variety of measures of human behavior, are are pretty accurate and pretty, you know, accurate to a point of being, useful and predictive. Right? So there’s a whole variety of use cases where AI has been wonderfully efficient and effective, long before the sort of generative AI revolution. Which I think has actually impacted a lot more of the quote unquote verbal aspect of the conversation than the nonverbal. So the ability to understand the words in the conversation, has improved greatly, with some of these new technologies, but the nonverbal aspect has been leveraging a different approach from a technical perspective that’s been pretty powerful for at least the last 10 years. Totally. Yeah.

Rachel Cossar: And and by nonverbal, are you talking more about, like, the, like, the vocal components of speech or are you also talking about, like, body language and and those behaviors. Yeah. All of it.

Skyler Place, Ph.D: So eye contact, head motion, leaning, nodding, tone of voice, intonation, speaking rate, everything except for the meaning of the words, if you will. Right. Yep. Yep. Totally.

Rachel Cossar: It’s so interesting because obviously that’s that’s been our big focus at virtual Sapiens since we started and, the the generative AI revolution, I like that you call called it that, was so interesting because all of a sudden, it became table stakes to have the words also be analyzed because it also was just, like, so easy now through APIs to, like, loop these things in. But the nonverbal behavioral stuff’s been a lot more interesting, I think, at least for us to to explore because we do we do leverage actually on client on device. Compute power. And so the real time component for us has been, like, it hasn’t immediately broken the bank. Like, we haven’t had to raise tens of 1,000,000 of dollars just to get out the door, which has been great. But it’s it’s just fascinating to see things continue to evolve in this space. Yeah. No. Absolutely.

Skyler Place, Ph.D: And it’s it’s become interesting because the you now have the optionality as a product designer to decide which stream of information you wanna access, right? Yeah. And, and especially, you know, in in near real time, you can now get the words and the conversation with low latency, along with the nonverbals. So you get to decide which of these information streams are most valuable for the use cases that you’re that you’re supporting. And that’s a big change from how things were a couple years ago. Yeah. Right. Definitely.

Rachel Cossar: And what are your thoughts on, because, you know, there’s the application of using this technology as a a coach, right, to receive feedback on yourself. And then there’s the kind of everyone’s infatuation with being able to, like, read the room and decode people’s because you said something earlier that I think is really interesting and and and important and and it’s that there’s a a difference a little bit between someone’s intended message or their, like, the intention behind their communication and then how they actually come across Right? Like, that’s Absolutely. Always at the root of all of it. So if you’re using AI in real time on a video call, for example, to deduce what other people might be thinking or feeling. Mhmm. But it’s not like how do you, like, how do you reconcile that with the fact that, like, you you can’t know their intention. Mhmm. Yeah. Yeah.

Skyler Place, Ph.D: And it gets even more there’s another layer of complication with this because there’s the goals of the individual in the conversation. Right? So Yeah. It’s like I have a worldview of how I’m thinking. I have a worldview of how I’m presenting myself I have a perspective on how you’re receiving this conversation, and then I have a goal in mind of what I’m trying to communicate. Right? And what’s really interesting when you think about an AI coach is which of those sort of personas are you coaching on? Right? Yeah. And and what are you training the AI on? In terms of the truth because the quote unquote facts of the conversation may be totally irrelevant to the perception of how things are going. Right? And and if and if, if the person on the other side of the conversation perceives the conversation is going really well, it doesn’t actually matter if if you think you’re stumbling through everything. Right? So in some sense, like, what’s what’s best for an AI coach? Is it Yeah. The facts of, your speaking rate is higher than ideal in a conversation like this. It’s like you’re coming across as nervous. Or is it your body language is presenting in such a way that you come across as confident and that is a stronger signal then your speaking rate being too quickly of being nervous. And therefore, the person on the other end is actually receiving this as a an expert who knows what they’re talking about. Right? And in which of those sort of aspects of a conversation. Yeah, shouldn’t AI coach be be focused on? And that’s the really interesting part of inserting another entity into a conversation. Yeah. When humans are used to human interactions, right, and do you build an AI that is, that interacts with you in a way a human does, or do you purposely build an AI that is focused on other aspects of the conversation that a human isn’t as aware of as the conversation is happening. Right. Totally. Does that make sense? Yeah. Totally. Yeah.

Rachel Cossar: I mean, and and because we’ve had to put a lot of thought into this, obviously, a virtual safety issue. And, like, our view on it is if our AI can be a mirror that we hold up to say, like, these were the behaviors you exhibited on the call. We like to leave it up to the user to to integrate and to reflect on the feedback that they’ve received and whether or not it aligns with those goals and those intentions. Mhmm. Right? Like, here here’s the way you’re showing up here as some of the, perception based like communication trends that those might affect. It like, is that where you were going for? Because if it wasn’t, then you know, then then you can start to maybe think about changing certain behaviors in certain contexts. And then the other piece that we do is we really do encourage a human coach to be in the loop so that the human coach can help with some of that, like, synthesis. Right? Yep. Absolutely. Yeah.

Skyler Place, Ph.D: It’s been it’s been interesting to see sort of the the branding and product development in the space in the last couple of years because the a lot of the companies in the early days, quote, unquote, were branding this type of approach as emotion AI, and it was about recognizing emotions that humans were presenting in a conversation. And I’m a big fan of the fact that the industry has moved towards understanding human behavior. Right? And it’s much more about what what is the what is happening in the interaction? What are the behaviors and leaving the the human to be the expert to decide what are the emotions that I am trying to express or that I’m feeling within this interaction?

Rachel Cossar: I’m so glad you brought that up because that’s been such a big, evolution for us as well. Like, from the beginning, I I I was very reticent to use any of the AI that had already been developed in terms of like, oh, detecting emotions and, like, this is the expression you’re making blah blah because I was like, there’s such a difference. There’s such a leap for the expressed emotion or the emotion detected for lack of a better term and then the felt sense of emotion. Right? And And I was like, the last thing we’re gonna do also is tell someone, like, that they should be smiling more. Or, you know, and so so we actually we don’t, reflect back any emotions. We just look at we’ll give you, like, a facial expression variation reading. And so you can you can see, oh, my my facial expressions varying a lot in this part of the conversation and then it went completely flat or or neutral or stayed happy or, you know, it’s it’s up to you to develop that awareness, but but we can reflect back on the variation just not on the emotion. Right? Exactly. Exactly.

Skyler Place, Ph.D: And it gets back to that question of, you know, what are the strengths that the human is bringing to the conversation and how do you leverage those as a AI tool builder so that you’re not distracting, folks with misinformation relating to how they’re, quote, unquote, supposed to be feeling which may not be actionable or particularly useful in the moment. Right? Totally. Totally.

Rachel Cossar: Man, I could keep I’ll I’ll go to the next question, but I I’ve talked to you about this whole day. So in terms of these applications of it, I like, what are you most excited about in terms of the continued evolution moving forward. Yeah.

Skyler Place, Ph.D: So I think what I’m excited about broadly is the availability of AI coaches for pretty much every role in the workplace, particularly in tech companies, but across a much broader set of industries as well. And I think the, you know, the the OpenAI chat GBT, the attention they’ve received has made this approach much more of, the awareness in the in the common non expert, if you will, has has increased dramatically. So folks are asking the question, you know, how can this help me with my with my job? How can this make my, you know, my life at home, more efficient and more effective? And I think you were seeing a little bit more openness to folks who may think that their skill set is to specialized or to experience driven to have a tool that could help them. And I think everyone from the CEO, you know, down to the intern is thinking, how can I, do my job more effectively and more efficiently and leverage some of these new technologies? So I think the idea of an AI coach for every knowledge worker is a reality that’s in front of us and how to, you know, ethically and appropriately implement those approaches are where, as a consultant, I spend a lot of my time working with organizations, helping answer some of those questions. Yeah.

Rachel Cossar: What has been the most, like, surprising, role, but that’s been that AI has been able to help coach that you’ve seen. Yeah.

Skyler Place, Ph.D: So I think the ones that I I’m, least knowledgeable about and therefore most surprised that has been in the design space. So so thinking about, UX, UI, graphic designers, Adobe and some other organizations have really come out the next generation set of tools to help support the the creative, professions. And and that’s a, a market where there’s in particular a little bit of in my experience, hesitancy of, you know, does my unique skill set as a designer going to be replaced in this idea of augmenting instead of replacing. There’s there’s been some some really great conversations I’ve been having with folks on how to leverage these tools to make your client, you know, be able to work do do better work for your clients, not just faster work. And and and that’s been a really interesting evolution of how some of these creative roles are thinking about using tools to augment their abilities. So. Right. Yeah.

Rachel Cossar: We, just moved my, my husband and I and our our son just moved and we have a gallery wall of art and my husband was trying to find a program that could help him within, like, the AI space of, like, combining virtual reality or it’s like AR and and then getting, was so funny trying to get the AI to give us different dimensions and different designs of where we could put our pictures on this wall. It really was not it was really having a hard time. So I don’t know if maybe there’s, like, it’s still a little bit a lot more work within the actual, like, physical design space. Right. Still, Yeah. Yeah. No.

Skyler Place, Ph.D: I mean, and as you know, this the technology is changing day, day by day, week by week. So I was talking to a few a few friends who are in the interior design business and trying to figure out if there are tools that allow them to do, you know, better visualization of furniture and spaces because Right. Relevant, you know, relative to what you were just describing and and they and they haven’t there’s no there’s no silver bullet yet. I think for that, Yeah. But I think it’s an area where there’s been rapid improvement in the last, you know, days weeks months. Every every time I talk about this, I have to, you know, re re equate myself with the state of the art because it it’s changing so quickly. Totally. Totally.

Rachel Cossar: And and you’ve mentioned a few things in this regard. What are you most concerned about as we continue to push the boundaries in this space? Yeah.

Skyler Place, Ph.D: So I think I mentioned earlier, one one of the things that I was really excited about was that was that business leaders were pausing and asking the question of what’s the appropriate use case. And I think that on the flip side is also I think the biggest concern is There’s a lot of, attitudes and tech about, you know, just try it, break stuff. Let’s see what happens and, you know, pick up the pieces and move forward. You know, keeping the pace of innovation very quick. And I think when we’re talking about tools that are changing how how people’s jobs are being done, how people’s performance in their jobs are being measured, when the actual human experience of work is changing so dramatically, we really have to think, deeply about the appropriate use cases of these technologies. And I think today, the ability to coach and expert is a great use case for some of these generative AI approaches. So as long as you know enough about the topic area, the hallucinations, the fallacies, the misinformation, you as a human can weed out the truth, from the from the misinformation. But if you blindly start applying these solutions to replace human decision making or to replace critical entire roles. They aren’t ready for it. Right? And and you’re gonna run into many, many, many problems across the spectrum. I mean, I think we’ve all seen the news, the the, you know, the the kind of the joke of the of the week around, Google recommended adding glue to your pizza recipes. Yeah. Right. But, you know, imagine that as it relates to performance reviews, marginalized populations. Right? Like, the the information that’s being fed into these is a representation of everything we’ve ever written or consumed on the internet. And as we all know, the internet is not always right on every topic. Right? Yeah. So that’s that’s I think the the ethics around this, the the boundaries of use cases, having business leaders really stop and think and bring their employees into the conversation. And really think of the use of AI in the same way you would evaluate new product features for your users. So, you know, having doing user research, talking with your employees, really understanding, what is helping, what is hurting, and and what enables them to be successful Yeah. As compared to looking to the top line and say, oh, if I bring on board an AI coach, I can reduce my cost by 10%. That’s my only definition of success. The human element of this, I think, remains absolutely critical. Totally.

Rachel Cossar: I one of the things that I’ve like to think could come from this, like, rapid adoption of AI is, is a little bit of that pause and thinking critically about the information that we’re being given. And I think that that goes for anything that we may ask of and receive from an from an AI bot or or assistant, but then also, like, you know, reading the news. Like, There’s the the the disinformation, excuse me, that has been around for, like, many years. This is not a new thing. Right? But maybe now we’ve come to see just how powerful that the access to information that might not be fully accurate is and and maybe as people, we can really start to develop more of that, those those critical thinking skills. Yeah. No. Absolutely.

Skyler Place, Ph.D: And I think it’s it’s even more important relative to the non verbal AI solutions. So if you think about, you know, what we were talking about in terms of measuring behavior versus measuring emotion or intentionality, if you can imagine, you know, flash, you know, fast forward 5 years and, you know, part of your performance review is the body language that as a sales leader, you exhibit in front of clients. Right? Yeah. And you’re being evaluated on on your communication style automatically my AI solution that the sales leader has brought in because they think that getting everyone to behave the same way in front of the clients will lead to a more consistent selling motions and higher higher sales rates. Right? Yeah. But, you know, when you have and, you know, if your behavior changes, the question is why. Right? And is that because of things happening in your personal life? Is that because of the previous conversation you had? What part of that is ethically appropriate for your company to be aware of or to be evaluating. And when you start moving from using these tools help the employee Yeah. To using them to blindly evaluate the employee. Yep. That’s this really dangerous line, I think, and it’s particularly so within the the nonverbal space. Totally. This divide between observe behavior and hidden intentionality. Yep. Yep. Absolutely.

Rachel Cossar: And there was one other thing I was gonna say about that, man, just escaped me. But, yeah, I think the, making sure that that that that’s what it was, that that there’s not too much of a prescript prescriptive nature either to the feedback I think is a delicate balance because you don’t want everyone to be getting the impression that, yeah, like you said, we want cookie cutter Mhmm. People with cookie cutter behaviors. Yep. Right? So it it’s so fascinating, though, like, the level of nuance and and and thought that has to go into any kind of application of AI in space, but just just to know that it’s possible is exciting. Yeah. No. Absolutely. Absolutely.

Skyler Place, Ph.D: And there are, you know, there are companies today that are selling solutions that, you know, if you follow this script, you will sell more because this script is what the top selling individuals in your organization are doing. And, yes, on average, you know, you look across a big data set, the statistics are true, but the ability of an individual to understand in the moments what is the right response and right way to deliver that response is a decision tree that has many branches. Right? Yeah. And so there is no one right answer in the moment. And I think that once again, we get into how does an AI help coach versus what are the strengths of a human in the interaction and that ability in the moment to know, you know, what is the right next step to move the conversation forward I think it’s something we wanna preserve and and coach on not try to average out across across everyone. Totally. Absolutely.

Rachel Cossar: Well, Skyler, thank you so much for the time to to show up and to share so much of your thought in this space, you know, to many people, it feels like AI has only just now come on to the scene, but as you know, that’s not the case. So Anything else you’d like to share with our audience? No. I think it’s an it’s an exciting time.

Skyler Place, Ph.D: You know, AI has come from, you know, niche use cases to something that we all, I think, are thinking about on the daily. And I would just say, you know, continue to think and continue to make good decisions for you and your organization. And, there are experts out there to help, if you if you need help. Awesome. And and to that point, how can people reach out to you? What’s the best way?

Rachel Cossar: Yeah. So I’m I’m on LinkedIn. I’m just Skyler Place on LinkedIn.

Skyler Place, Ph.D: I run Skyler Place Consulting, and I work with organizations of all sizes to help them figure out how to take advantage of some of these modern AI tools help move their business and their employees into a better position. Awesome. Great.

Rachel Cossar: Thank you again, and thanks, of course, everyone for tuning in. We’ll see you next time. Wonderful. Thanks, Rachel.