Generative AI in the Workplace

Transcript:

Pete Wright:
Hello everybody and welcome to Connected Knowledge from Upland Software on TruStory FM. I’m Pete Wright. Today, we delve into the world of generative AI and its far-reaching impact on our work.
Sure, we’re seeing all sorts of cutting-edge tools like Copilot and Azure OpenAI Services and Google’s Vertex AI and Duet. But how are these technologies reshaping the way we work and interact? Join us as we explore the next steps in the AI revolution and the cultural shifts necessary to adapt to this uncertain new world.
Skailar Hage is a product marketing specialist here at Upland, and he’s with me to talk about the future of work and the role of generative AI in shaping our nine-to-five and beyond.
Skailar, welcome. First timer, good to have you.

Skailar Hage:
Awesome to be here, Pete. Super excited to dive into this topic.

Pete Wright:
Are you watching the Olympics at all? I’m dating us right now.

Skailar Hage:
I’m a huge surfer, and I love the basketball as well, so I spent most of this weekend watching both the men’s and women’s surfing and then also watching the men’s team beat Serbia yesterday too. So that was cool.

Pete Wright:
That’s super fun. And what you will notice as an avid Olympics viewer this year is Microsoft is all in-

Skailar Hage:
Everywhere.

Pete Wright:
… and they just want me to watch them. That’s the tagline. Watch me. They want me to watch. Copilot is everywhere. So I wonder, as we kick off this conversation, addressing this idea of how AI is impacting the world of work and the culture at work that we’re seeing such a huge spend to get people to think real hard about it at home on such a big stage as the Olympics, that’s probably a knife that cuts both ways. What do you think?

Skailar Hage:
Yeah, I think AI is special in this… Every 10, 15 years there seems to be some new technological revolution that everyone says is going to change everything. Internet 2.0 was big with social media, which it did. And now we’re kind of entering into the 3.0 space.
And one of the great things about AI is that no matter you’re a large enterprise or you’re just somebody who runs a mom and pop business running Excel sheets, doing QuickBooks, whatever it is, copilots and all these different technologies are going to affect you in one way or the other.
So I think it’s wild to see the big broad target market that Microsoft is going after, whether it’s ads in the Olympics, ads on LinkedIn that they’re trying to target for C-level executives at these large enterprises. It’s wild. All of it’s crazy.

Pete Wright:
Well, and everybody’s so beautiful, handsome that are using these tools. I mean, the best looking people that you’ve ever seen at work, they are asking you to watch them collate presentations. And it’s very exciting and that there’s room to be optimistic.
So let’s talk about what you’re seeing in terms of how… I think, what are we on? Two years really since ChatGPT?

Skailar Hage:
Just about.

Pete Wright:
Yeah. So where do we stand right now if we’re charting the long arc of history or the short arc of history for these AI tools?

Skailar Hage:
Yeah, that’s a great thing to talk about. I think right now we’re kind of peaking in my estimation. So they talk about hype cycles. There’s this massive curve that goes straight up. It’s almost a straight line up into enthusiasm, awareness of the technology. It starts to level out, and then it kind of just drops instantly. But then over time it kind of equals out.
And I think we’re kind of descending towards where the midline would be eventually. I think a lot of organizations over the past two years have gotten caught up into the hype, and maybe a lot of enterprises didn’t really think through exactly how they’re going to enable AI. And they threw a bunch of money at it, and now they’re like, “Where’s the return on investment for the billions of dollars we’ve just put into these projects?”
So I think we’re going to kind of get towards the eventual mean here in the next two to three years. But there’s been a lot of money that’s been dumped into these AI projects. A lot of organizations are seeing successes from it. But at the same time, something like Copilot, how can you really measure… How can you get a number from that, like a return on investment number? It’s just improving productivity.
So it’s something very, very hard to gauge. There’s no metric that you can really set to help fuel that more investment into this. So although it is helping productivity, it’s very hard to measure from the executive level is what I’m seeing.

Pete Wright:
I would challenge to see what you think about how you measure it on the employee level. In some respects, are we seeing adoption to the extent that we aspire to in the Microsoft Copilot ads? Are we seeing people change the way they work individually to the point that we can start measuring anything?

Skailar Hage:
I think that’s a great point to bring up as well. And every organization’s different because a lot of organizations are very innovative. They have this innovative culture, move fast, break things, adopt the best technology possible, and then we also have laggards, we call them in the hype cycle, that maybe they have money to throw at this type of technology, but the culture doesn’t allow them to truly utilize it in a way to the way it’s meant to be used.
And I think those are the organizations that are really going to get caught behind. Because if you put a bunch of money, resources, into buying the technology, having people learn about it, and then also having the training from Microsoft or whoever, consultants, whoever might be, and you’re not seeing that success almost instantly, that’s a very, very big, scary risk that these organizations are taking.

Pete Wright:
I remember, maybe because I’m a little older than you, I remember a time when computers shipped without a spell check, and then spell check was a third party add-on that you’d have to buy.

Skailar Hage:
Right.

Pete Wright:
And we interviewed somebody on this show last year whose forecast was there will be a time when AI is like spell check. That everything ships with AI somehow as a part of it, and you won’t know when you are using it or not. It will just be running in the background. How close do you think we are to that point?

Skailar Hage:
In my guesstimation, I would say about maybe five to 10 years, I still really do think it is new and a lot of organizations are definitely scared of it. Everybody, you go on LinkedIn or you see these ads that you’re talking about on the Olympics, everyone’s very gung-ho about AI. “Oh, yeah, we’re AI enabled. We use AI for XYZ.”
But at least what we’re seeing is that organizations are still very hesitant because of, one, security. It’s a very, very large risk, especially in highly sensitive industries like finance, healthcare, legal, those types of industries where things are very safeguarded.
And it’s going to be a tough, I’d say mountain to climb, but then once we’re there, it’ll absolutely explode. I think that’s really when it comes, when those laggards and those sensitive industries in the next three, five, maybe even 10 years, start to really catch on. I think that’s when everything will just be, like you said, shipped with AI.
We won’t even talk about it. It’ll just be a part of things. It’ll be, I kind of liken it to what Apple released with their partnership with OpenAI. So the next iPhone is going to have just AI embedded into it.

Pete Wright:
Apple Intelligence is what we’re talking about. Yeah.

Skailar Hage:
Yep. Which is a great branding by the way.

Pete Wright:
Yeah, AI.

Skailar Hage:
Apple intelligence. I loved that, by the way, the marketing.

Pete Wright:
Let’s just say it better work.

Skailar Hage:
But that’s what makes Apple great with things is that they kind of sit back and watch and then they perfect whatever they’re doing. That always seems to be their story, but that’s a whole nother piece of things, but-

Pete Wright:
Yeah, we’ll have you back, and you and I can nerd out on Apple stuff.

Skailar Hage:
For sure, for sure.

Pete Wright:
I want to say, because you’re talking about security as one of the major concerns, and I wonder what you’re hearing in organizations about ethics and ethical considerations around using AI. And I’m talking specifically about training data, about displacement of jobs and retraining, what are you hearing about those concerns? Are those bubbling up at the organizational level at the C-suite level, or is AI just being shipped with our big apps and it’s just integrated?

Skailar Hage:
No, absolutely. I mean, security is huge. I mean, that’s the first thing somebody, especially an executive, thinks about is what’s the security aspects of this? And one thing that we’ve really tried to enable is not only the security, but the accessibility of information as well.
Because when things are as technical as they are with implementing generative AI, the accessibility factor still need to be there as well. So those users still need to access whatever knowledge that they need to do their jobs. And it’s kind of a fine-tuned balance when you have to be extremely secure in the information you’re safeguarding, but also have it accessible to those who need it to further productivity. Does that make sense right there?

Pete Wright:
Sure.

Skailar Hage:
Tip toe.

Pete Wright:
Yeah, I hear that. And I feel like that’s one of those things, like every time I read something about AI, it’s, “Look at the incredible promise if you figure out how to use this stuff right.” And on the other side is the creative unions. It’s the people who fear their jobs will be displaced by these things by copyright being a legal platform to actually sue these companies for training data.
And I think there’s a sweet spot in here, which is in organizations training their own AI models on their own data. And big institutions have a real opportunity to both leverage the, I’ll say it, magic of the technology at the end-user level and maintain the sort of integrity of product.

Skailar Hage:
Yeah. I get where you’re coming from, and I think this also brings into the evolution of where we’re going with this AI in terms of, instead of these large, massive LLMs, large language models, that we have that take these vast amounts of data, we’re really going to start becoming specialized in turning into SLMs, which are these small language models.
So for your example, maybe this is a massive enterprise we’re talking about with a ton of different types of data. Instead of deploying one massive large language model for every single type of worker they have, whether it’s IT, marketing, product, et cetera, they can have specialized models for not only their function, but whatever tasks that they’re trying to accomplish as well.

Pete Wright:
Sure. Like call center support gets their own. Right.

Skailar Hage:
Right. And I think that’s where things are really going to transform because it’s not as generalized. I’m sure you’ve played around with ChatGPT, Claude-

Pete Wright:
For sure.

Skailar Hage:
… whatever model it is. A lot of the times it’s not very good of an output. But when you give it that highly specific data where it’s personalized and I guess personalized and accessible to you, then that’s when things really start to change because then you have all of the data that you need. It knows what you want to hear, how you speak, how you type even. That’s where things really start to make an impact.

Pete Wright:
I’m an AI optimist.

Skailar Hage:
I am. Me too.

Pete Wright:
Right. I think we’ll figure it out. But I don’t think any conversation on AI right now is complete unless we at least say the quiet parts out loud-

Skailar Hage:
For sure.

Pete Wright:
… and it’s hard. So let’s talk about next steps, and I’m really interested to hear what comes as organizations do normalize into using AI. What are the sorts of things we can expect on the horizon two years out, five years out? What do you see?

Skailar Hage:
Great question because I’ve been thinking about this a lot the past month or so. I personally think that the next evolution of this is going to be sharing the information that we get from these models, whether your colleagues, your superiors, whoever it might be.
I think it’s going to be extremely valuable to have multiple models of these running. So say you have three different models that are running. You have some sort of input and the three models give you three different outputs. You see which one of those are the best, or something can kind of aggregate that information and give you the best output possible, and then you can share that information.
I think when the knowledge that is created from these models can be shareable and dispersed throughout your team, I think that’s when things become very, very valuable. Because not only does it increase the collaborative effort, but it’s also going to over time better the output of these models as well. I think that’s where I see the evolution headed from a, I guess, technical and workplace perspective.

Pete Wright:
How will increasing sophistication of generative AI of these models, and sophistication of small language models too, how do you think it changes the way we perceive and interact with our technology at work for productivity purposes?

Skailar Hage:
Well, I mean, the more sophisticated these models get, in theory, the better the output should be. So, in theory again, our productivity should go up. But at the same time, talking about the dark side, how we were earlier a bit, the more technical and complex things get, the more need for maintenance, the more scrutiny of security there is, the more training data is needed for these said models.
So as we become more sophisticated, things either need to become a lot cheaper very quickly, or things need to become a lot better very quickly as well. Which I think, to our point, in the next five to 10 years, we’ll probably get to a point where things like GPUs and the cost of running a model will be probably next to zero in the grand scheme of things. So it won’t be that expensive to deploy highly specialized models that take a bunch of energy, but-

Pete Wright:
Yeah, right now AI is moving natural gas markets, so we kind of need it to come down.

Skailar Hage:
It’s crazy. Yeah, I mean, that’s what people don’t really think about too is the energy perspective of all these things too. I mean, the electricity alone is just unbelievable. And that’s a whole nother cost that most organizations maybe they’re thinking about, but not maybe to the point of they need to.

Pete Wright:
Yeah, right. Let’s talk a little bit about what you’re working on and what you’re excited about at Upland. What’s going on, and when can I play with it?

Skailar Hage:
Absolutely. So at Upland, we’re doing a lot of great things with AI. Almost every single one of our products in our KCM BU has taken a super, super innovative approach with AI. We’ve been one of the first organizations to kind of set up this AI board, as we like to call it, where we have designated people from all over the organization, whether it’s product, marketing, IT, we kind of all oversee what we’re doing with AI in our products, gives us a very diverse set of eyes to look at things and improve things over time.
I’m personally excited about the product I work on, BA Insight. We’re doing a lot of cool things with AI enablement. Speaking of security before too, that’s a huge part of this. And that’s one thing we’re very focused on moving forward is keeping security measures, not only intact, but improving them as we improve data to feed into these AI models.
But we got a lot of cool things going on at Upland, not just with my products, but across all of the business units.

Pete Wright:
It’s very cool and very cool to see these things evolve and actually become features of substance. So many of the AI tools where, as you were talking about, as the market just kind of exploded, they’re really just plugging in ChatGPT and OpenAI stuff. It’s just a window.

Skailar Hage:
It’s just a ChatGPT wrapper, you know?

Pete Wright:
Right. Right. And so, the idea of being able to see these more significant applications, as now we’re two years down the road, start to come out and really see what we can do, there is some real power in what we can do, and BA Insight is the perfect place to do it, so excited to hear about that.
On a personal level, what’s your personal AI stack look like? What are the tools you’re using and love?

Skailar Hage:
Kind of going back to what I was doing before, this is kind of where I got the whole knowledge sharing aspect, I jump from model to model to basically figure out what the best output is. And then I kind of, whatever I’m asking it, I’ll take the three outcomes I get and then kind of make my own spin of it.
Obviously, I’m the type of nerd that I’ll just literally talk to AI for like 20 minutes about something where I just want to, “Hey, say it this way. No, use it this way. Use XYZ input and give me blah, blah, blah. Change it to the way I write,” that type of thing.
But my, I guess, tech stack is just the basics like everyone else is using. I guess the only thing on top that I like is for content, there’s Jasper, things like that that are really well functioning, help you get in the right mind frame of writing content.
Claude’s a great model, OpenAIs got great models. And I also just saw that Meta is releasing an open sourced model of Llama too, so that should be really exciting to see where that goes as well. But I think everybody should at least have the basic… Whether it’s Claude, ChatGPT, whatever it is, somebody should definitely be very well versed in how to use it because it’s unbelievable what it’s done to productivity, just for even the basic package. You don’t even have to pay for the enterprise package or whatever it might be.

Pete Wright:
Yeah, yeah. I’ll tell you, I think the biggest secret for me in AI tools is called Poe, and it’s a service and app from Quora. And it includes all of the models, so you-

Skailar Hage:
That’s awesome.

Pete Wright:
It’s like a chat bot that you just open a new chat with the model you want to talk to today, and it’s got all of the new ones. It’s got all the Claude 3 bots through Opus. It’s got all the ChatGPT bots, it’s got Dall-E. It’s got Llama. It’s got Google’s bots. It’s got the whole suite, and you can create your own bot.
And I find you get to know the personality of these bots as you use them and realize that with creative writing and brainstorming, Claude is really better. But for checking my regex, yeah, it’s going to be OpenAI, and it’s going to be ChatGPT.

Skailar Hage:
That’s funny you say that because I think Claude is the best for copy, and I like GPT for data analysis. That seems to be the thing that I stick to with both.

Pete Wright:
The other one that I’m playing with right now, just since we’re talking about personal stack is Perplexity AI. Where do you stand on Perplexity? Have you tried it?

Skailar Hage:
So I’m not super familiar with it, but I do know of the name. That is something I should look into, but give me the skinny on it. Why do you love it?

Pete Wright:
Oh, I love that I just threw down the AI trump card to the AI guy. This is delightful. Delightful for me. No, Perplexity, it’s a little bit controversial, so let’s get the controversy out of the way.
It was one of the first bots that was looking at all of the websites across the internet and ignoring robots.txt. So it was scraping data without permission and without compensation.

Skailar Hage:
Okay.

Pete Wright:
So I started using it and testing it before that news came out as we… I’d say probably four or five weeks ago. But it has a really interesting model, which is to give you, you ask complicated questions, and it gives you essentially a brief on your answer with retraceable links that are as thorough as any model that I’ve ever seen. It is as close to the answer engine aspiration that I’ve seen yet.
And so, for asking complex, multi-step questions and getting essentially an outline of responses with all the click-through URLs to take you back to the source sites to verify, it’s a pretty special tool.
To your point specifically, though, and why I’m bringing it up, it actually has a tool within it where you can say, “Create a webpage with this collected result.” So your whole conversation is similar to ChatGPT, but you can create a really beautiful, well-designed, essentially research page to share with others. And, like you said, the more we get toward sharing our work, the better.

Skailar Hage:
Absolutely. It’s funny, I was writing a blog recently about all the different copilots and AI tools that are available, and Perplexity is one that has been popped up, but it just hasn’t been on my list of the countless AI tools to dive into.

Pete Wright:
There are just a few. Put it on the list. It’s fun to play with and interesting. Interesting model.

Skailar Hage:
For sure. And going back to what we were saying too, so I think, speaking of talking back to the ethical part of AI, it’s super interesting. Because we get to this point where, I’m an AI optimist, like you said, I want AI to be used to improve human. Everyone always uses the human in the loop phrase. It’s super big right now.
And I want it to improve my productivity. I don’t want it to displace or replace anything. In my personal life, I’m super creative. I love all things creative. I won’t dive super deep into it, but I love using Dall-E, for example, for image generation. I love, “Make an alligator swimming in the ocean,” or whatever it might be, some different prompt.
And what I’m finding is a lot of the first step that we’re going to is replacing people like creatives with AI because it’s like, “Oh, I can just have an image generated of exactly what I need. I don’t need to have any type of artistic ability anymore.”
So I hope that… The optimist in me says that we can focus technologies on improving productivity and enhancing our abilities instead of just total replacement. But that’s a whole other conversation we could, again, dive into is the ramifications of everything.

Pete Wright:
As an AI optimist myself, I think you’re right on. And what I have to keep telling myself is I am not a designer and a graphic artist because I don’t have the taste to know whether the goofy image I just created is good or not. It really is just a goofy image. That’s why we hire designers. They know better.

Skailar Hage:
Exactly. My father’s a graphic designer, so it’s like I get torn between it. But at the same time, I could also take that same methodology that I spoke about before. It’s we can enhance everybody. But I just think a lot of the times people gun for the arts and the humanities and things, when in reality we need to enhance those abilities.
And those skills are going to become very valuable as AI explodes because a lot of technical skills are going to be displaced or replaced by AI. Things like coding, coding, for example, it can spit out code in five seconds that maybe a dev team couldn’t do in six hours.
So I think it’s all going to be about using that artistic and human ability to really help tune the models and then also enhance what we do overall.

Pete Wright:
Skailar, it’s been great. Thank you so much for coming and chatting with me about this stuff. I feel like we have more to talk about, so I hope this isn’t the last time. Yeah, right.

Skailar Hage:
As long as you’re happy to have me back, I’d love to come back and chat it up.

Pete Wright:
Absolutely. This was great. Thank you so much. Where can people find you? Should we send them to your LinkedIn? Any resources you want to connect folks to?

Skailar Hage:
Yeah, absolutely. Follow me, connect with me on LinkedIn. Always happy to grow my network. I’m super active on there, whether it’s writing blogs, resharing some of our content here at Upland, all the cool AI stuff that we’re doing across all of our business units. Probably you can go to my page and see all the stuff we’re doing. So, yeah, I would just point everybody to my LinkedIn page.

Pete Wright:
Excellent. Well, we’ll put links to your LinkedIn. We’ll have links to BA Insight, some of the tools that we mentioned personally, I’ll put links in there too. They’re fun to check out. Enjoy them. And thank you everybody for downloading and listening to the show. We appreciate your time and your attention.
We’d love to hear what you think. Swipe up in your show notes and look for the feedback link to send questions to us or any of our past guests, and I’ll do my best to get them answered on a future show.
On behalf of Skailar Hage, I’m Pete Wright. We’ll see you right back here next time on Connected Knowledge.