By Use Cases
In this episode, Dr. Natalie Petouhoff, from Abundance360 joins host TJ, VP – Product Marketing at Yellow.ai to discuss how combining AI with human empathy elevates customer experiences, and explore the value of empathy, trust, privacy, and the potential of large language models – a blend that drives the future of fully autonomous customer interactions.
Intro- 00:00:03: Generative AI takes the center stage. But is your enterprise still watching from the sidelines? Come on in, let’s fix that. This is Not Another Bot: The Generative AI Show, where we unpack and help you understand the rapidly evolving space of conversational experiences and the technology behind it all. Here is your host, TJ.
TJ – 00:00:26: Hello and welcome to Not Another Bot: The Generative AI Show. I’m your host TJ. Joining me today is Natalie, a very well known name in the industry. Natalie’s impressive career traverses the world of technology, customer and employee experience and business transformation. As a renowned business consultant, she helps brands reduce costs and increase revenue by transforming the customer and employee experience. She has held key positions and consulted for companies like Salesforce, Hulu, Marriott, and General Electric, among others. And currently, she’s making significant contributions as an Abundance360 member and consultant, as well as the Chief Customer Experience Officer Consultant and expert in residence at The Customer and Employee Experience Advisory. Natalie is also the co-author of the bestselling book, Empathy in Action: How to Deliver Great Customer Experience at Scale, and I’m super proud to have one with me. With a profound insights into the role of AI in shaping customer experience, and her deep commitment to embedding empathy into technology, she’s a true visionary in our field. Natalie’s mantra, that is, “what is good for employees and customers is ultimately what is also good for companies, forms the backbone of our work”. Welcome Natalie, we’re pumped to have you here.
Dr. Natalie Petouhoff – 00:01:37: Thank you so much. I really appreciate it.
TJ – 00:01:41: Awesome. All right, Natalie, let’s get started. Can you share a bit more about your journey from studying methodological engineering to becoming a recognized leader in customer experience in AI? It’s a complete shift in terms of going all the way from engineering to building something more specific to Customer Experience, Domain and AI. We’d love to understand how that shift happened and if you want to share a little bit more insights there.
Dr. Natalie Petouhoff – 00:02:02: I come from a family of engineers. My dad was an engineer and he took me to work with him at Ford Motor Company. And I met a woman who was looking at the microscopic properties that then led to their ability to be used in a car. So why is a bumper made out of the kind of material it is and why is a tire made out of material that can withstand going over big holes and all kinds of things. And so to me, it was really fascinating to look at the connection between the structure and how it performed and then the results. And so as a young engineer, I was often asked to consult on projects and I saw a lot of the suggestions that were made weren’t really customer centric. So there was a hole in the book I talk about, an opportunity to galvanize steel and for the car doors and the decision got made that was more along the lines of planned obsolescence. And so I realized that I had this knack for understanding the customer and also the employee because that decision made an impression on me as an employee. So I realized I really had this orientation, maybe it’s just in my DNA to be thinking about the customer and then also how does the work dynamic impact the employee and then how does that connect to what we deliver for customers and employees in the long run. And it’s just been an interesting, so now I’ve just basically, I mean, there were a series of events that happened, but I ended up getting hired by PricewaterhouseCoopers to consult in the call center space and CRM. And so yeah, it’s been an interesting journey and I’ve used those engineering skills to do root cause analysis and start to really connect some of the dots, especially when it comes to customer and employee experience.
TJ – 00:04:04: Wow, that’s awesome. I mean, this massive shift, but also we know that the sort of experience and expertise you bring and things you do is just amazing. And especially for us in this field who are building platform customer experience, it’s just a massive validation when we hear from you and the sort of learnings, it’s totally empathic. So awesome. Good to hear. Well, before we go into the actual conversation a bit on the AI and generative AI, I do think that we also want to learn, and I’m sure the audience would love to learn about what inspired your book. I mean, some of it I think you mentioned already, but Empathy in Action is a big thing itself. So what inspired your book, Empathy in Action, and how does it tie into your overall mission of your heading from here on?
Dr. Natalie Petouhoff – 00:04:47: So I think the concept of empathy really came from my early days working in the auto industry. And one of the people that I really studied was Edward Deming. And so if you remember that story, he had been trying to get the attention of the American auto industry to really look at the feedback from customers and employees. And they basically weren’t open to that. And so he ended up going to Japan and had the ear of many executives and leaders there. And what I saw was, so if we define empathy as the ability to hear the other person’s perspective or point of view. What we’re really doing is we’re collecting feedback. And oftentimes feedback where we may have a blind spot. Or we may just not know what we don’t know. And what I find is most people see the world through their lens, through their eyes, right? Their perspective, and it totally makes sense. And where the richness in life comes from, getting that other perspective. And so we have the ability, we’ve always had the ability to talk to customers and talk to employees and look at that data, but it’s really a mindset perspective of how valuable that is. And so for me, the concept of empathy means that you’re actually listening, understanding what your customers and employees want and need, and then integrating that feedback back in to what you’re doing so that you’re delivering products and services that really meet the needs of your two most important assets.
TJ – 00:06:28: Absolutely.
Dr. Natalie Petouhoff – 00:06:29: What’s fascinating to me is that it seems like common sense and yet we have this perspective that we know better. And I’m not pointing fingers, but it’s just like we as human beings think we know better. And what I’ve learned in the process of becoming what I would call empathy practitioner is that there are a lot of things that all of us, we don’t know that we don’t know. And having that other perspective can really change the choices and decisions that you make. And I think there is no field that this is more applicable to than what we’re doing, especially when it comes to technology.
TJ – 00:07:06: Totally, who also rightly said, the opinions are always around, but I think you definitely have to get that outside in view and lead with empathy to kind of even understand what customers and employees are saying. So very well articulated. Now, keeping this topic to what you just mentioned, now, when it comes to automating customer experiences and customer service or support, like companies like us who are doing that, question to you would be, what does empathy really mean to you in that sort of set up? And how does this differ from human-to-human interaction? What is leading towards this particular question is that, how do we predominantly translate human empathy into an AI language and keep the empathy still kind of intact while conversing with a chatbot or AI bot or automated customer platforms?
Dr. Natalie Petouhoff – 00:07:56: So I think the first thing is the decision, whether you’re a software developer, a vendor, or whether you’re a company implementing this, is are you really customer and employee centric? And so what I find is people think they are until I start to ask them questions and they realize, “oh, we didn’t really see that we weren’t”. And so really deciding that and really understanding whether you are or not. And then I think another thing that’s really important is to understand what is it that you’re trying to accomplish. So if we just look at the technology itself, can the technology feel? No, but can you design an experience that emulates the feeling that you’re really talking to a person or that person gets what we’re asking for? And the great thing about the large language models is that it gives us context that we’ve never had before. And that context is really what creates the feeling of relatableness, I believe. And so when you start to think about what is it that you’re really doing and delivering an experience, you’re trying to take the input of the customer. And take and deliver a solution that’s going to satisfy their needs. And I think chatbots historical and self-service in general was limited. And so now we have a trust problem. Because if you use the chatbots we all have in our personal lives, oftentimes we’re met with, I’m not really sure if the chatbots is going to give me the answer I want, or I got an answer that didn’t make sense. And so now we have to rebuild that trust. What’s great about platforms like ChatGPT and all of the others is that it has the ability to build in this contextual model that would allow us to respond in a way that did deliver those kind of interactions. But now we’re going to have to train customers to want to use it, because otherwise what they do is they may use self-service, but then you can see that same customer in your first contact resolution. If you’re keeping track of the number of times someone contacted someone, they call the agent just to check. So what I think is really exciting is that we have this opportunity to use technology that can do so much more than we’ve ever had before. And at the same time, I think that we really need to be mindful of how we’re designing it. And part of that, I think, is that because we have so much, so I think one of the things, that we had talked about looking at is the ability to really become customer service. So you might say, “well, how do I know?” So we have so much data in the contact center from texts, emails, voice. So if you take that and you do what I believe is pronounced concision, which is synthesizing and creating a concise summary, you can look at the volumes of data very, very quickly and be able to get a perspective on what’s working and what’s not working. And once you had that information, now you can really get a picture of current state and the gaps and really solving for what didn’t work. And so that’s a way of using a large language model to become customer-centric. And then once you have that information, then start to design the product or then start to design the experience. And I know everybody who’s in technology wants to jump straight to, let’s see what the thing can do, right? Let’s test it, let’s play with it. I mean, I’ve had a lot of fun with ChatGPT myself in terms of asking a question and honing my prompt skills and really trying to figure out what its capabilities are and it’s really cool. But I can see where oftentimes we just jumped to the technology and we haven’t really considered what the customer really wants. And so you had asked me about customer satisfaction or loyalty. It’s really about designing it. If the customer gets what they want and need, have a marketing question. If you have just a general question or a sales question or a broke fix kind of question, if you can really help that person get the answer, now we’ve created someone that feels satisfied and that satisfaction is really what’s going to drive the loyalty. So what’s fascinating to me is that a lot of the principles that we’ve had in this field for a very long time still apply and they apply even more because if you think at the speed at which we’re going, if you didn’t know what you didn’t know, you could be delivering the wrong things faster.
TJ – 00:12:53: Well, you just kind of hit the nail on the head. It’s just certainly have the expertise to explain it, but explain it so well. For us, we have been building these experiences on the platform for the longest time. Certainly building very enterprise specific large language models, combining it with the CDPs, the knowledge search. We have a very specific multi-lelem architecture, which is certainly coming out well. I think biggest thing, and we’ll talk more about it in this conversation too, is how we were able to create smaller models to be very specific on the use-cases. I think a lot of the times we’re talking to customers because they all want to kind of jump into this whole technology innovation. It’s like, “hey, take all of it and just put it in generative AI”. But I think we are going with the idea. What is your problem? What’s your use case? And then we have built these very specific large language models for summarization, for document Q and A, for cognition, and a lot more of those. And we’re constantly building for FAQ, which is basically helping and being very specific to what customer really wants for the use-cases and then deploying it for them. So I think to your point, building experience and then getting and thinking about building loyalty when they’re satisfied is critical, but also the fact that how they’re thinking about the adoption from a use case perspective is equally critical. On that note, what would you say Natalie is the most significant obstacle when trying to make AI more empathetic? And how can businesses really navigate this? Because to your point, it’s new, it’s very human-like, with large language models, with NLP innovation and generative AI. But then there’s still businesses who want to navigate further and they really want to feel that human-like interaction. Are there any proven strategies to overcome any of these current obstacles to make AI more empathetic? And if you have any examples of a use case you want to share, it would be great.
Dr. Natalie Petouhoff – 00:14:47: I think it really does go back to we as the humans control the machine, at least for now. I know there’s a lot of people that are worried about that whole concept. I think that if we really think about how are we going to create, let’s say, a personalized experience. So Mercedes-Benz just released a chatbot for their car and their app. And what they did there was they’re using voice prompts. And some of the issues there were around trust and privacy. So if we start to think about using some sort of walled garden approach, so rather than just taking all our customer data and putting it into the OpenAI system, what Mercedes has done is use Azure OpenAI services. So that they’re using Microsoft’s Cloud and the AI platform within a specific instance. And they’re calling it the Mercedes-Benz Intelligent Cloud. So there, you’re overcoming the issues with privacy and trust, which are really key, because if people don’t trust it, then they’re not going to use it. And now we start to really think about, “okay, so if we ingest a lot of customer data and the types of questions that customers are asking, and we have the context, we understand what’s prompting them to react, and then we have used the database of all the information and the reasons why, now we can start to contextualize when someone says these things, and before bots were very rule-based, and they didn’t really have the technology to understand context”. And so what I want to say is, there’s not magical thinking here, right? It isn’t magic. I mean, there’s still technology that’s crunching things, right? So part of what we need to understand is, what you feed the model and how you train the model, especially in a walled garden situation, is very key because that’s where it’s going to generate the interactions from. A lot of people say, “well, how do I get ready?” Well, part of it is, just looking at your knowledge base, looking at your workflow processes, looking at your previous chat transcription, and starting to really… And so what’s cool about this is the ability to synthesize all that and to come up with root causes and ways to look at why do people call. And so to me, being empathetic is you understand me as a customer, you have pre-emptively understood 80% of your calls and there’s always going to be some percentage of interactions that are going to be outside that norm. But you really have studied as a company or as a vendor the reasons why I need help. And then you’re designing that experience with the data to be highly predictive. So you could use predictive analytics to be able to create a really great interaction based on my preferences, based on the things that I’ve done before, looking at patterns and the next best action. But again, that comes from integrating some of that historical data. And so the data is still really important because it’s not just some magical thing that the bot is really, really smart. It does require training and that’s part of the,
TJ – 00:18:18: It’s awesome. Yeah, I think one of the things with it, Natalie, I mean, I think you will be a little bit happy to hear about this is also down. So we have been using most of the vendors in our space, have been using large language models for some time because otherwise you certainly can’t build a context or the sort of outcomes which are important for complex situations. I think the biggest thing we have seen with the adoption of net new large language models and also our own large language models we have built is to solve for the complex situations and bringing a lot of the human-like interactions and empathy, which was not possible. And we were able to now touch a long tail of the customer situations and use-cases. But I think the biggest thing for us at this moment, what we’re definitely doing is this whole thing or concept around goal-based conversations, Natalie, where we have created basically a dynamic chat window in our platform. So rather than in a bot building framework, you would have to go through the entire workflow, write every intent, every utterance. And then train them. So one and a half years back, we launched something called as a DynamicNLP™ , which was a pretrained model, like all 12 billion conversations that happens on the backend, anonymized data, certainly. And then that helped in kind of going to the next step with the LOG, which is our GPT based Large language model, to bring that human-like interaction. And we’re seeing close to, 80 to 90% automation and self-serve because of the flows are created dynamically based on the prompt that’s written. My question to you is, it may be a little bit futuristic, but I’m sure you touched upon it to an extent. Would you have any sort of thought around a fully autonomous customer experience world and down the line two years from now? Because we’ve seen around 80 to 90% for complex situations too, but for some use-cases, not all per se. But will there be a world in next one or two years or even more where you would say everything being fully autonomous, just like Tesla cars being full self-driving. Do you have any thoughts on that?
Dr. Natalie Petouhoff – 00:20:21: So there are really a couple schools of thought here. One is that AI is going to take over and it’s going to eliminate call centers. And I think there was when self-service first came out, it was the end of the contact center. And there was this concept of if you actually do everything right, they won’t call you. So I think that that’s not possible to have contacts. And I think also the idea of not having any contacts comes from thinking of the contact center as a call center. So if you think about it as a profit center or an opportunity to connect and build relationships and build loyalty, now it’s a relationship center. So you actually do want to interact with your customers because that’s where you’re going to drive the relationship and the trust and then the long-term loyalty. So starting to think about designing, why do people need to interact with the company? And then designing experiences that deliver things that they actually need. So one of the reasons, and it’s kind of a fascinating case study that the reasons Amazon became so successful is that they used a lot of customer advocacy groups. So every single day they would talk to customers who are using the platform and say, “what do you like? What don’t you like?” And they would get feedback and they would integrate that all the time in terms of really making a platform. If you Google and you look up pictures of what Amazon looked like years ago, like you can’t even imagine the marketplace that it is today. And the way that they really did that was they have this concept of when they have a meeting, they save a place in the room, a seat for the customer because everything is through that customer lens. And so I think that’s a really good lesson learned in terms of future-proofing and trying to really meet that need. Because again, we go back to what kind of experiences do we want to design? Who are they for, it can be kind of an intellectual fun, let’s see what the technology can do, or it can be, let’s really figure out with these people who are contacting us, what do they really need? And then using all that amazing engineering and brain power to be able to deliver an experience that’s focused on creating those experiences and using the relational models that have now context to be able to deliver an experience that is going to garner that trust so that people will use it. I still think that people will contact contact centers. I think whether it’s by phone or whether it’s by chat or email or social even, I don’t personally think but we’ll see, right? Jury’s out. I don’t think that you’ll ever have a time when you don’t need humans. And I think there’s, like in my own consulting that I’m doing with DoorDash, we find there’s a lot of instances for the type of cases. I mean, DoorDash is a delivery company. If you’re a Dasher, you’re looking for information about work, or if you’re a customer, you’re like wondering where your food or something that you got from the drug store. So it’s a delivery company. And what we find is there are some interactions that really work really, really well with respect to AI and to being able to deliver self-service or very, very minimal people interactions. But let’s say that you’re a Dasher and you are in the midst of delivering something and you walk up to someone’s house and you get bit by a dog. So now can self-service help that Dasher tell DoorDash that delivery person just got bit by a dog and they’re not going to be able to complete the delivery? And might the Dasher need help? And the way that we started to really look at different levels of detecting emotion, detecting indicators of frustration, looking at the need for human interactions came from really looking at thousands of chats and phone interactions and determining emotional indicators, right? Does someone repeat the story? You know, in the chat is there like WTF, bold, lots of question marks, right? So these are indicators. I’m frustrated. You’re not giving me what I need. So starting to look at that historically when does that happen? What kind of situations or use cases or scenarios does that happen and how can we better be more proactive about anticipating those so that more things can be automated. And then there is a certain percentage and every business is going to be different. So I think that there are situations that you can look at to really determine when you really need a human interaction and when can something be fully automated. And so if it’s an emergency or it’s about money or payment or someone being sick, there are some situation, I think that technology may not be able to handle the emotional side of that interaction. And they really need to talk to a human being and be reassured that the companies got their back. So I think it comes from aggregating those types of experiences and situations and scenarios and really understanding them. And I think some businesses will be able to automate quite a bit and other businesses may not be able to automate as much.
TJ – 00:26:08: I think it just comes down to the use-case as well, like how complex it is and where there would be an engagement with the human no matter what. Totally agreed on that part. Now, I think we touched about the large language models as we were speaking, but as you look into the large language models like GPT-4, LLaMA, or many others like Bard, what potential do you see in their future creations? And if you may have a thought around what do you think the next advancements in this area might look like, purely to kind of enable custom experiences beyond.
Dr. Natalie Petouhoff – 00:26:42: I think that the ability to really look at the amount of data that we have and really start to ingest that data. I think part of the reason that we haven’t been more customer or employee centric is that in order to do that, so part of its executive mindset is that we want to do that and we really care. And this is a priority. And then the ease at which that customer feedback can be ingested and analyzed and given back in a concise summary so that now we can make strategic decisions. And we can start to model how does the change in those strategic decisions affect the bottom line? How does it change average handle time? How does it change FCR? So I think the connection is going to be the advancement of the models as they get smarter and smarter and the ability we’re training them and the more data that we give them, the more intelligent they become. And as we do that, I think part of this process to really pay attention to and where the experimentation and the challenge comes from is really paying attention to as we feed it that data, what didn’t we know that we didn’t know? And as the machines get smarter and smarter, we also need to be understanding what it’s revealing to us about what we didn’t know. I think where technology really becomes applicable is when we start to figure out how does it apply? So, so many times what I see in technology as a product in search of a solution, produce it because it’s cool. But why deep down, why do we care, right? And from a pure technical engineering geeky side, right? It’s cool. And it does something we’ve never been able to do before. And it’s fast, incredibly fast. I mean, part of my own personal realization of, wow, this is an exponential, not there’s a step change, but this is exponentially faster and smarter and better than anything we’ve ever had to date. So part of that is that personal realization. And then where I think, because a lot of people talk about technology taking our jobs, I think those of us in this field, where the real utilization of a traditional CX person or employee experience HR is going to be in understanding the technology, what it can do for us, and then beginning to extrapolate how it can be applied. And it’s in the application that I believe, not the magical thinking, but the magic is going to happen. And so for CEOs or CHROs, people who are looking at talent, I think we’re looking for a different kind of CX person. We’re looking for a different kind of engineer, right? We’re looking at someone who can help the machine be applied in a smarter way. And as we apply it in a smarter way, the machine gets smarter, but we have to keep up with what the machine is showing us it can do. And so I believe that it’s really going to change the nature of roles and work, and it’s going to change the nature of what it means to be in the customer experience world. Like if you’re not on ChatGPT, you’d be to your one of the models out there really need to jump on and ask it to plan a vacation to Italy and learn like you’ll hear this term prompt strategies. Someone that said, if you were advising CEOs, what do you think their biggest blind spots are around AI? And what advice would you give them? And so I thought, well, let me ask ChatGPT. And so I came back with some really interesting points. And so I was like, “okay, I hadn’t thought about that”. So then I was able to apply all the years of things I’ve learned about technology, about people, about organizational change, about resistance to change, about mindset, Perspective, leadership skills, and take the information that the machine gave me and then add the color, which is where the application comes from. So the machine can spit out a whole bunch of points, which are great. But now how do we take that and we make it real for someone else? And I think that’s the next level of advancement is really us learning from the machine and the machine learning from us. And when we take that and we really combine that, the more intelligent we are about using the machine, the better our prompts and our input and the way that we’re using it improves because we’re really training it. I mean, if you think about a washboard stomach, if you want a washboard stomach, then you have to do sit-ups. But if you don’t train, you’re not going to get there. So AI is not exactly like doing sit- ups, but. In the same way. We have to have some discipline and some doing it because we want an outcome. I mean, and at the same time, there’s things that we don’t know, right? So there is some aspect of large language models and the technology that is a bit of a BLACK BOX. And so depending on what model you’re using, there’s some things that we don’t know that we don’t know.
TJ – 00:32:18: Yeah, it’s just getting trained continuously to enhance the model itself so that it can give back better answers. I think that’s what, when I was talking about it, like we took the journey long time back with DynamicNLP™ and then built our own large language models. I think one of the key things we are able to deliver is this whole prompt-based dynamic chat where you just write the prompts. You don’t even drag and drop or you create a workflow. The workflows are created dynamically on its own based just on the prompt. So when we talk about this, analysts and others are like, “okay, how did you guys even go that far?” Because we were looking at the experiences and talking to customers a lot and we figured out the pattern that we kind of for specific industry, write prompts like these and then make it very use case-specific to the customer. We’re seeing tremendous amount of results that includes context. To your point, like if you’re planning a vacation in Florida, the dynamic chat is like so powerful and it’s out of the box, which means you spend zero time to set up. But you’re asking here, I need to find a hotel near the sea beach in Miami. So that’s first part of the context. So it shows you the hotel straight away. And then the second thing it says, “oh, but I’m also traveling with my dog”. And now the bot is able to now reply back because of this prompt-based dynamic chat that, “oh yeah, you can find these hotels near the beach in Miami, which is Pet Friendly”. So I think that’s where, you know, what you just said, I’m just trying to validate that with your experiences, if you’re saying so, some of the things we’re doing and I think customers are seeing the value straight away. They feel that experience has gone better. So I just want to add to that, but I’m glad that as a company, what we are trying to build is aligned to how you were trying to explain it. So, you know, this is really, really good to hear and big validation. On the second part, Natalie here is the, what advice do you generally give or have for businesses that are just beginning to incorporate AI, I mean, not only just generative AI, but AI into their customer service operations, especially in regards to maintaining a very empathetic approach? What are the success criteria they should look at and what sort of strategy they should build around so that, you know, they are ahead of the game with their competition?
Dr. Natalie Petouhoff – 00:34:33: I think, again, it goes back to what are you trying to accomplish? So if you’re trying to increase the ability to self-serve, so then you want to start to look at what kind of chatbots could you deliver that could really provide context? Some of that is the ability to do predictive analytics. So let’s say you’re interacting with a chatbot on a website, and let’s say you’re logged in, so you’re authenticated so they can see your previous history data and you’re kind of wandering around a website. You are shopping, then you put something in your shopping cart, and then now you’re looking at frequently asked questions. At some point in that customer journey map, if the technology can start to map where I am in the journey and it can see that I’ve now stopped my shopping experience and I’m going to ask questions, learning to figure out when do you pop up a chat, right? So there’s lots of times you go to a website and you land on the website and it says, “how can I help you?” Well, you’re like, “I don’t really need help yet because I haven’t figured out what I’m doing”. So timing for popping up a chat bots is really important. So it’s important to be helpful but it’s also important not to be annoying. And so looking at that customer journey and the customer journey analytics and then figuring out where in the process is it most helpful to offer a chat? And then can that chatbots take into consideration historical buying patterns? Is this something new? Where am I? What frequently asked question am I looking at? In the FAQs, what products am I looking at? And then the bot could come back and say, “hey Nat, are you looking at this particular product? Do you have a question? You’re looking at the FAQ”. So I think that kind of intelligence is now possible and really fun. And I think that’s how you’re going to build trust with customers is when the technology actually delivers something that the customer is, “oh, wow”, like kind of a wow experience. And that was really helpful. And even though there’s like some people I’ve talked to are like, isn’t that kind of creepy and big brotherish? In some ways it is, but I think in the end of the day, if it’s really helpful to the customer and it doesn’t feel like you’re being spied on so you can have some gigantic big upsell cross sell, and it’s more like maybe a little bit of a recommendation engine. If you like this, then you might take a look at that. Then I think again, it goes back to this concept of really serving the customer.
TJ – 00:37:17: Absolutely. Wow. You’re awesome, Natalie. I can just go on listening to it. Just like golden words to the ears. Music, absolutely. This is amazing. Okay, two more questions before we finish this. One is, I think you touched upon it a bit. What role do you see Generative AI playing in the future of work, particularly in customer service and support? And how can businesses really best prepare for this, especially if they have to stay ahead in the curve and the rapidly evolving field, it’s becoming very competitive for them as well. Customers we talk to, we definitely know that they’re here to solve for some, not only just the mundane tasks, but also complex scenarios on self-serve so that they can really get the customers excited and give them the wow experience and satisfaction. And then wherever it may be an escalation to the agent. Any thoughts on this precisely where it’s heading and how businesses can stay ahead keeping Generative AI in mind?
Dr. Natalie Petouhoff – 00:38:14: So I would say part of it is in terms of the employee experience. And I think this comes from the top CEO level, that the CEO becomes a champion of generative AI and says, we as a company are going to use this. And then I think because what I love about ChatGPT and Bard and those kinds of applications is because the user interface. Gives those of us who are not programmers the opportunity to interact with a large language model and generative AI in a non-technical situation. So it’s that for me, it was a visceral experience having its experiential, right? We can talk about the concept of generative AI, but it’s not until I feel you use it that you really understand, “oh my gosh, it created a marketing brief for me”, or “it wrote like points for my paper”. And it’s not like the human being doesn’t then have to take that and turn it into something that’s useful. But there are some productivity tasks that can really be made useful. So I would say in general, there needs to be some sort of decision that we are going down this path, we are embracing it. And whether that’s true or not may have a lot to do with where someone falls in the adoption curve. So is your leadership innovators and early adopters, or are they part of the late majority or like, we’re going to say, this is never going anywhere, and the internet and email didn’t happen. So there are people that are slow to adopt, and they have their reasons, and it is what it is. So depending on your outlook on where we are, and where we’re going, those of us who are innovators and early adopters have jumped in, so I would say, I would talk to the CHRO, I would figure out how is this changing the nature of work? How can we give our employees an experience of this? And having some sort of guidance or training, just to get that experiential knowledge or wisdom that does for a feeling of, okay, there is a there there, all these like people talking about it. And so maybe everybody has an account. And I would say those of us in technology probably have an easier time with some of this because we’re used to when some of the social networks first came out, we jumped on and we didn’t know how to use it either. But the cohort of people that jumped on co-created how it’s used. So I think it’s really important in the workplace to decide that you’re going to do this, and then have trainings like webinars explained to people like, “okay, how can you use Bard or ChatGPT plan a vacation or do something that and then have people experience that”. And then I think part of it is, in terms of employees, how can it really help them in their work? So maybe it’s productivity. I know firms like Korn Ferry are using it to do personalized job searches and job matching. They’re looking at for performance management and rewards. Companies like Salesforce are creating a talent marketplace. For current employees to match them with open roles and learning. So I think there’s ways to start to really incorporate it into the employee experience. I would also say that we’re also in a time where if you look at like Gallup or McKenzie, some of their most recent studies, they started to look at employee attitudes. And I think they said maybe 77% of employees are not engaged. So 23% are. So when you start to look at that and you start to look at the lack of engagement, and then you start to think about AI could be scary if people don’t really understand it, that might make them even more afraid and more distant from their job and less engaged. So having some sort of education, some sort of stance on here’s how we’re going to approach the future together is really important because those statistics don’t really incorporate attitudes towards AI. And in many cases, and in some cases, jobs will be changed a lot. And in some cases, they’re going to be eliminated. So if a job is, and this takes the executive level to think about this and the whole HR group to start to think about roles, responsibilities, and what’s the application of AI and generative AI to those roles and responsibilities all the way down from the CEO, all the way to a frontline employees, even like on a factory floor robotics, and then figuring out how is that going to change our workplace roles, responsibilities, and then what do we need to do for workforce planning? So how do we prepare? How do we reskill, upskill the workforce? And then how do we bring in generative AI tools and capabilities that are really going to enhance that employee experience? So if you have a population of employees that’s really not all that engaged, which if the statistics are true, if you’re paying a dollar for every piece of salary and you’re getting 23% engagement, there’s a cost factor. What that says to me is there’s a cultural aspect here. And I think part of what happened during the pandemic is it gave us a cause for pause. And it allowed us to really think about work and life and what we really want. And as a result, we asked ourselves some really big questions and we said, “what do we want out of life?” And so I think that we’re in the midst of a zeitgeist where people have a very different opinion and very different ideas about what they want from work. And so I think unless executives are really tuned into this, they might go into generative AI thinking everybody’s on board when either they’re fearful or they’re resistant or they just don’t really care. And so when you start to think about how can you turn that fear or resistance into possibility and excitement, it’s those firms that are really looking at this employee equation from how do we take those natural emotions, which we tend to just discard. We just like say, get on the train, get on the bus. If you don’t get on the bus, it’s going to leave without you. And there are some CEOs that would say the bus is going to run over you. So I don’t think we’re at a place and time where we can just ignore employee feedback. I think we really have to understand that. And part of being able to get good feedback is creating a psychologically safe place where people can say, I’m scared, I’m frustrated, I don’t understand, raise their hand and be able to truly express those things, maybe have dissenting opinions. Because there are some people that are like, generative AI is the best thing since apple pie. And then other people are thinking, you know, this is like the end of the world. And being able to be with all those different types of perspectives, understand it again, going back to the empathy equation is really starting to look at from that person’s point of view or this employee population’s point of view, this is how they feel about that. And then how is we as leaders in a company going to start to respond to that and really begin to understand are the things like I know that there’s a lot of tools around sales enablement. And I think some of them are really cool because the idea is to take all your salespeople and turn them into have them have the same characteristics and capabilities as your top salespeople, which in concept is great, but we’re almost turning professional salespeople into contact center agents because we’re measuring the heck out of them. And then we’re giving them constant feedback and have we prepared them to see this as the opportunity. To really improve what they do, make more money, have bigger commissions, have a larger share of the deals in the marketplace, or does it feel like they’re being spied upon? I mean, quite honestly, I was part of the days when ERP and CRM were implemented. And there are like if you Google historical failed ERP and CRM, there’s great articles around hundreds of millions of dollars that were spent on projects like this that went upside down. I mean, there were many reasons. Part of it was the technology was limited, proprietary systems, we over customized it and we couldn’t maintain upgrades and all that kind of stuff. But a lot of it had to do with organizational change and not really understanding and dealing with employee resistance. And so, I mean, I know I was on a site where we were ready to go live in somebody the night before going live for a can of code. Into the server. It was really painful, but they just felt like this technology had destroyed their personal and professional life over the past year. And this was a way of saying, I don’t really appreciate this. So what’s fascinating to me is we’re at that juxtaposition, that tipping point of really cool new stuff. And this really cool new stuff idea is not new, right? Every so often, like if you look at the last 100, 150 years over the progress of the Industrial Revolution, we keep hitting upon new areas of technology, exponential change. And at the same time, we really need to deal with the human element as part of that. What I’m really hoping, especially in co-authoring a book like Empathy in Action, is that we’re starting to really recognize the value of incorporating those feelings and feedback into, whether it’s customer or employee experiences, and then into the development and the deployment of technology.
TJ – 00:48:29: Just brilliant. I mean, very important. And I think that’s what we have been leading with how do we build human-like interactions or human empathy into the experiences? It’s possible, but I think there’s a lot of porosity to it. And I think it’ll take its own time to get there. We’re seeing some benefits, but the more we are learning from customers, I think we’ll also start implementing quite a bit of that. But Natalie, you have no clue how insightful this was. I’ve been reading the book. There’s so many great pointers there. But just hearing from you, the explanation of a few things and how you feel about the future and what every business should be considering as a strategy is so well-explained. And I think it’s going to be amazing to our viewers just listening in. So on that note, Natalie, thank you so much for being here on Not Another Bot: The General AI Show. We totally loved having you here and hope to speak to you again in the future. This has just been eye-opening in few regards. And I will send music to the ears to just hear your point of view, being an expert in the field. So thanks for the opportunity to have you here.
Dr. Natalie Petouhoff – 00:49:32: Thank you.
TJ – 00:49:33: Let’s play. How impactful was that episode? Not Another Bot: The Generative AI Show, is brought to you by Yellow.ai. To find out more about Yellow.ai and how you can transform your business with AI-powered automation visit yellow.ai. And then make sure to search for The Generative AI Show in Apple Podcasts, Spotify and Google Podcasts or anywhere else podcasts are found. Make sure to click subscribe so you don’t miss any future episodes. On behalf of the team here at Yellow.ai. Thank you for listening.