Amanda Razani: Hello, I’m Amanda Razani with Techstrong.ai, and I’m excited to be here today with Carm Taglienti. He is the chief data officer for Insight. How are you doing today?
Carm Taglienti: I am doing great, and thank you so much for inviting me to speak on the show today.
Amanda Razani: Glad to have you here. Can you share a little bit about Insight and what services do you provide?
Carm Taglienti: Yeah, so Insight is a solutions integrator, and so what that really means is that we work with customers, including our partner network, to be able to provide technology solutions to organizations to help drive their business and to allow them to be able to create competitive advantage or operational efficiencies within their own business.
Amanda Razani: Great. So today’s topic of discussion is generative AI and other AI tools and how they affect the customer experience. So I’m going to go straight into my first question, which is, how are generative AI tools currently used in customer experience environments today?
Carm Taglienti: Yeah, so that’s a great question, because I think in a lot of ways people think about generative AI as a productivity enhancer, if you will. And so if we look at it from productivity enhancements, you’d say, “Well, I can generate code faster,” for example. But as it relates to customer experience, I think it really is probably two things that I want to talk about. One is, how do I understand my customers better so that I know what kinds of things may improve the customer experience? And so there’s a lot that can be done with respect to being able to analyze information through the generative techniques where you can “talk to your data” through the generative experience and understand your customers a little bit better so that you can anticipate what it is that they would like to see or how they can interact with your programs or with your services.
And the second one is that generative AI also is the… I’ll use the image side of the house, where we can create images that represent or appeal to our customers. And so we can think about look and feel, and we can do that quickly. And so the productivity side of that also allows us to be able to do things like visual design. How can we create an experience that might be more amenable to the kinds of products and services that we might be creating for our users? So I think in that way, there’s really just sort of the art of the possible is facilitated by this generative AI experience.
Amanda Razani: So what industries or markets are utilizing this technology the most?
Carm Taglienti: So we’ve been seeing a lot of uptick in retail is number one. I think the retail industry is using it quite a bit for production of marketing content, for example, or even changing the way that they design their websites, or even just understanding their customer base more effectively. Sort of the examples I was giving you just a little bit earlier. But we also see it being used in the financial services again. Or interestingly, this is a popular one, is in the legal area as well. So understanding contracts and policies and then how can you use this information to know is maybe… Well, insurance is another one, I guess, kind of in a legal sense, but it’s sort of the, are specific claims within my policy? Or if there’s a particular lawsuit, where can I reference the kinds of statutes, for example, that might exist where it might apply?
And so it’s really just an amazing way to absorb a ton of information and then to be able to use that in an effective way. And so I think any industry that gets involved in that is going to use it in an effective way. But I think the most popular ones I mentioned. Oh, and healthcare I think, is another one. We’re not quite at that point where we see it being used for diagnoses, so we have to be careful with that one, of course. But certainly it’s being used in that context as well. So just sort of understanding patterns and maybe helping to ascertain what are some of the recommendations as an assistant, if you will, for a doctor or other kinds of advice that might be provided.
Amanda Razani: Are there any new generative AI tools or regular AI tools on the market that maybe enterprises should be aware of today?
Carm Taglienti: So in the context of large language models, and so what drives all of this is large language models behind the scenes, which are basically just… It’s a neural network that has a corpus of knowledge that was trained, or sorry, trained by a corpus of knowledge, but it’s focused on specific types of domains. And so where the information is collected from allows you to be able to do better or worse at specific types of tasks. And there are a plethora of different models that are out there. And as a matter of fact, you can even create your own. And one of the things I think that is becoming most popular is taking sort of a template model even from the open source community and then training it with your own data and then using it for your own specific purpose. And so in terms of, I’ll call it, new models or new capabilities, that’s a big one. And so customization for your own subject domain is becoming a very popular way to use these kinds of models.
The other one is, I’ll call it… It’s not necessarily new, but the technique is that we’re seeing what are called intelligent applications. How do I build this directly into my application? And we were talking about user experience before, so how do I create a dynamic feedback loop so I can enhance your experience, for example, by giving you the kinds of things that might appeal to you, for example, when you’re in my website? And I can do that by using this, we like to call it a copilot build in, if you will, but it’s sort of a copilot that works with your application to enhance the experiences. So these are really just new and innovative techniques, leveraging that underlying large language model that can help you to be successful. And I think those are really amazing innovations when you think about how this all works. And so those things are really new, and I think that probably over time we’ll see more advances in things like image generation as well as other characteristics of how we can use these generative AI capabilities.
Amanda Razani: When it comes to implementation of these tools, what is your advice to organizations that are just trying to harness this technology? Where do they start?
Carm Taglienti: So I think the… What I recommend to customers is, first of all, start to get an understanding of exactly what your use case might be. Sometimes organizations are like, “I need generative AI, I’m going to go whatever. I’m going to get an account.” Azure is an easy way to do it today, for example. “I’m going to get OpenAI running in Azure.” But then they don’t really have a use case for it. So you can play with it, but then you don’t really learn much. So the first thing to do, I think, is really just understand generative AI, what it is, and then start experimenting. And I think that’s really a good way to start. And there are many ways that you can experiment. Most organizations, though, want a secure tenant because you don’t want to experiment with secure data, for example, or things that you want to expose publicly.
So a lot of organizations will work with us, and it just so happens that we have offerings which allow you to be able to do that to start off quickly, where you can just start chatting with your own data, and it makes it simple for you to do. And that’s a good place to start, but it really is this experimentation to start. And then from there, I like to call it the generative AI lifecycle, but it’s understand your use cases, understand how to practically use it as part of this learning experience, and then figure out how to productionalize it so that you can move it forward. But I think that’s really where we start. And the beauty of generative AI today, and I know you’ve used it, and probably all of our audience here has used it as well, it’s easy to use. You can just start using it immediately and you learn as you go. And I think that’s really the amazing thing about it. It’s completely agile and you can just… As you learn, you start to think about new innovative ways to use it.
Amanda Razani: It really does open so many doors because of the user-friendly nature of it. So we talked about the starting point. Are there any pitfalls or roadblocks on these journeys that you see that you can share, and how can they be avoided?
Carm Taglienti: Yeah, that’s a great question. And I think some of the pitfalls certainly… Well, number one is really just understanding the use cases and the policy that you might have within an organization. So many organizations, it may feel innocent. You might say, “Oh, I’m just going to use my corporate information or corporate IP or maybe personal information from within my organization,” or even your own information. And you make it available out through, I’ll call it, the public OpenAI services, just one example. So that’s one pitfall. It’s like you may not be aware of the fact that you’re producing what’s called data leakage. You might be sharing information that you don’t want to share. So number one, that could be an issue. And many organizations start with policies to define the appropriate use.
I think another pitfall is really just understanding then what this specific use case is, and I alluded to this earlier. It’s one thing to play around in the environment, not really understand what the specific outcome that expected is. And I think… So you can get caught up in a science project. And so a lot of organizations sort of get caught up in, I’m playing with it, we’re not quite sure how to turn it into something that we can really use to benefit the organization. So I think focusing also on return on investment, and we help customers understand how to do that. So it’s more of the, “What problem you’re trying to solve, what’s the actual return on investment?” And then it’s sort of strategically how do you get there and so that you can start to realize benefit from it? Because I am probably using generative AI a large part of my day. Most of it is productive, some of it’s not. But for an organization, you really want to think about how do you use it productively and not just play around in the general sense, because sometimes that happens.
So I think that’s a pitfall, too. It’s sort of understand what your real objective is and then sort of focus on that objective. And so that’s just another area. And maybe the last one is true, productionalization, and we saw this with general AI, where before you were able to say take a data model and release it and use it effectively we didn’t really understand what the production processes were. And so what happens is that someone in the data science team, you wanted to run a model to project potentially what are our sales for the next five quarters or what’s the likelihood we should invest in a particular product, and they had to run it on their desktop because we didn’t really have a way to productionalize it, if you will. And so now we could see the same thing here. So you get people sort of experimenting, and they turn into production without putting in the discipline processes around things like DevOps or pipeline support or even measuring and monitoring the performance of the model. Those things are another pitfall as well to make sure that you can get there.
Amanda Razani: This technology is advancing very rapidly. It feels like it just came on the market, and I know it’s been around for years being developed, but it feels like it just came on the market like the past year and just exploded and now everyone is talking about it. So where do you see the world in a year from now as far as AI technology use?
Carm Taglienti: That’s another great question. And if I knew that, I would probably retire. No, I’m just kidding. But I do see sort of the continued adoption of these capabilities and I also see innovation. And so there’s a lot of organizations that are thinking of new and different ways to be able to take advantage of this and how it evolves. And I also see this evolution of a framework approach. So it’s easy to use today. I think it gets even easier in the future. And so there’s an entire ecosystem of people that are building out plugins or add-ins or applications that allow us to make this even easier. So I think that’s where things go over time. I also think that there’s a bit of concern, and I think we need to see some regulation as it applies to sharing of personal information or use for personal purposes or even how to secure the environments more effectively.
So I think over the next year, six months to a year, we’ll start to see hopefully a little bit more work being done on responsible AI and how organizations use this effectively, because I think it’s really just… It’s a bit of a gold rush right now. Everybody’s going off and trying to figure out, how do we use this? And it’s great, but I think we have to take a step back a little bit and understand what are the implications to society and to… And I don’t want to go too far a fold on that one, but in general, what are the implications over time?
Amanda Razani: Absolutely. So is there any key point that you want our audience to leave with today or any additional tips or guidelines you’d like to share?
Carm Taglienti: I would say embrace the capability. And I think the other thing that’s really important for us to understand as just a community is that think about how it can augment your ability to be good at what you do. Because if you think of it as really a personal assistant, then it can help you to do your job better. So how can you think out of the box in terms of what can gen AI do to allow me to be able to go faster, to be more creative, to be able to do more or to do it more effectively? And that really, I think, is probably the most important thing that I would say for people. Don’t worry too much about it taking your job, I think, because in essence, you’re augmenting what it is that you can do and the effectiveness that you can have. And I think it just makes us all better in the grand scheme of things.
And I know it’s a very positive outlook on it, and there are some of the negative ones as well. But in general, I think it is important to think about it that way because I’ve seen it. I’m sure many people in your audience have seen it as well. And that’s really, I think, a key focus. But I also do want to say the other thing to keep in mind is also think about what are the practical use cases and don’t turn a blind eye to the fact that if we have these amazing capabilities, there are the malicious users out there that also have that same capability. So we shouldn’t get caught up too much in the fact that it’s only being used for good, because just like cybersecurity came to the fore over the last decade, the same thing is possibly going to happen here. We just have to be careful in terms of anticipating if it’s deepfakes for visuals or for speech. Just be aware that those things can be done, and we should all just sort of raise our awareness to those kinds of things as well.
Amanda Razani: Definitely. It will be interesting to see what the future unfolds. Thank you so much for coming on our show today and sharing your insights with us.
Carm Taglienti: My pleasure. Thank you for having me.