Can We Trust ChatGPT? Going Down the Rabbit Hole of the Implications of AI

Joseph Drambarean: Welcome to “Fintech Corner”! I’m Joseph Drambarean. I’m your host and the Chief Product Officer here at Trovata. And we are so excited to be joined by Kevin Bell, our Head of Client Solutions.

I’m so excited about today’s conversation because I’ve been meaning to pick your brain just anyways on the client journey, how we navigate just through the entire process just purchasing our software, what it means to bring it to life in a way that brings you value as a customer and… I’ve been curious to just kind of get your thoughts on generally what’s been going on in the marketplace, especially within tech. And so I just thought this would be a fun conversation. So thank you, Kevin. Welcome! I appreciate you being on. Second guest. I’m so excited to have you.

Kevin Bell: Yeah, absolutely. It’s quite an honor, and it does really feel like I’m just sitting here in the living room with you.

Down the ChatGPT Rabbit Hole

Joseph Drambarean: It’s funny. In the last episode, we touched a little bit on ChatGPT – and you had been mentioning it to me, and I was so tempted to go down the rabbit hole. And since we’re talking about innovation and… We were talking about how, if we’re using a basketball analogy, taking the baby steps of being able to make a layup and being able to do the basics right. I’d say that ChatGPT and advanced AI solutions – that’s like effectively the NBA Jam, jumping 30 feet in the air, backflip-dunking. That’s the most advanced use case there could be.

First of all, I’m just curious, have you been reading or keeping up to date on what’s been happening with ChatGPT, being the big reason?

Kevin Bell: I think you’d have to be hiding under a rock to not have heard about it by now. Yeah, I was joking that… A month or two ago, I told my parents who are both in their 70s about this. And they might have been the first senior citizens to be informed, but their mind was blown just as mine was. So that was kind of one of those moments where you’re like, “Wow, this is a pretty gigantic shift in the tech landscape.”

Joseph Drambarean: Yeah.

Kevin Bell: The iPhone coming out or even the internet being a thing in the first place are some of the only parallels I can really think of. It’s just like the democratization of AI across so many different use cases. So it’s pretty exciting…

Joseph Drambarean: Yeah.

Kevin Bell: …to think about the possibilities out there.

Joseph Drambarean: The tech part of me is elated. I am eating it up. I’m obviously a user. I spent hours with it, just like everybody else, tried to find the corners, discovered the same things that people have found.

So, one of the things that has been interesting to me about ChatGPT is the ability to take it down a rabbit hole, steering it and seeing the responses that you’ll get out of the interface. And it got me thinking, “What if…?” I’m not saying that this is true, but hypothetically. We’ve been talking about all of the basics and having a great software footing on having a good grasp of your cash, then taking it to the next level, being able to categorize forecasts, all of that. Well, what if you remove the need for having great software and it’s just a question-answer? You could ask, “Hey, should I be investing right now?”

Kevin Bell: Yeah.

Joseph Drambarean: Tell me. Given where I’m at from a cash perspective, where the money is going, give me a recommendation: How much of my cash should I be putting into an investment? Or am I gonna have enough money three weeks from now? Should I be thinking about making changes? How do you…? Do you feel that’s a scary outcome: a future where it’s not a stretch at all…that that could be a thing in the future?

Kevin Bell: No, I honestly really don’t, and let me tell you why. I guess part of the reason is I’ve seen, in past experiences with Salesforce and Tableau, that concept of conversational search…

Joseph Drambarean: Right.

Kevin Bell: …be kind of implemented and be used by organizations to really… Okay, you don’t need to know every bell and whistle of this BI tool or this analytics solution. It’s just whoever you are, whatever your role is within the organization, you can ask a simple question and try to get the answer that you’re looking for. So, to me, that actually makes a heck of a lot of sense going forward. And now, with the power of AI to support that even more, it could be a really interesting topic to play with.

Joseph Drambarean: Do you know what I’m afraid of, though? I’m fully in the same boat as you. I’m excited. I wanna see it. I wanna play with it. I want that possibility to be a reality. 

The Fears of ChatGPT Automation 

Joseph Drambarean: But then, I think about my mom, her first experience with ChatGPT. My brother is as big of a tech nerd as I am, so… He’s younger than me, and so he had a chance to sit down with mom and just expose her to this.

And the fear layers there are, of course, number one, “Will I be irrelevant?” Can we automate to such a degree where the inferences that a solution like this is driving are making the strategic part of the job totally irrelevant because, at the end of the day, you’re never gonna be able to compete with a computer when it comes to math, right? And a lot of what is done here is various layers of analysis that are based on “if this … then that” scenarios that just do math. So there’s that part of it.

Can We Trust ChatGPT?

Joseph D: The second part, though – and this one, I’ve been thinking a lot about just in the general context of ChatGPT, and I’m curious what your take is on this – is, “Can we trust it?” Because ChatGPT has shown so many versions of answers that are mutated from the original source where… Because it’s extrapolating, distilling, trying to take the best of Wikipedia and articles that it finds over here and a website that it’s scraped over there and summarizing, it is doing it using modeling. And so, there’s no guarantee that it got it just right. The only guarantee is that refinement over time will get it to a percentage point of satisfaction where, today, maybe it’s 99.99, but in the future, it might be 99.9999999 and that would mean that you could really trust it.

And what’s funny about just watching the interactions online is finding those corners and how adamant it will be that it is correct and keep steering down that path. And it makes you wonder, could you imagine scenarios where, as a business, you’re asking a ChatGPT agent, “Hey, should I make this decision based on what you know about my cash?”, and it takes you down this course – and it’s just flat-out wrong.

Kevin Bell: Yeah, and you’re making critical business decisions based on…

Joseph Drambarean: Yeah.

Kevin Bell: …that wrongness. Yeah, well, I thought you were gonna say you worried about the machines taking over. So, at least, we’re a step down…

Joseph Drambarean: Well…

Kevin Bell: …from that. Is that something you’re worried about?

Joseph Drambarean: I didn’t wanna turn this podcast into a conspiracy theory about AI, but yeah. Yeah, obviously, I have read all the stories about the Bing search falling in love with the end user and convincing them to do weird things. But yeah. That being kind of like the most extreme case, I guess the one that is right in front of us is, “Can we trust this technology, and what…?” That’s a more general question…

Kevin Bell: Yeah.

Using AI in a Creative Process 

Joseph Drambarean: …about AI, ML, its role in our use of tools and in the creative process of getting to a strategy or to an inference. And this is what we spend a lot of time debating, to be honest…

Kevin Bell: Yeah.

Joseph Drambarean: …here at Trovata. So I’m just curious what your take is on it.

Kevin Bell: No, I mean, I think it actually kind of proves the point of retaining the need for the human aspect of this solution to help verify and interpret and use that…

Joseph Drambarean: Yeah.

Kevin Bell: …critical thinking that, at least at this point, AI is not advanced enough to be able to do. And I think a lot of people would welcome… Hey, if I have – let’s just call it – a really, really incredibly powerful computer here that allows me to focus on the things that I wanna be doing, and I can use this as a tool to do my job more effectively and help my company achieve those outcomes more smoothly, I think most people would welcome that.

But at the same time, I mean, I think it certainly is going to warrant a lot of conversations and governance and guidelines, similar to how we’ve seen social media take off and then become…

Joseph Drambarean: Yeah.

Kevin Bell: …kind of a hotbed of some of the issues with fake news. Dare, I even use that term.

Joseph Drambarean: Yeah.

Kevin Bell: So I think we’re gonna go through a lot of that same chaos and confusion as we really understand how this technology unfolds.

Joseph Drambarean: Yeah, I think the mantra that we’ve tried to have for the longest time is, “Context matters.” Having the context of how the data was derived, what is happening is how a human plays an important role together with machine learning and AI, and that’s the thing that always makes me worried about… What I’ve seen so far is that it’s stripping a lot of the context, right? Of course, you can ask it, “Hey, what was your source?”, and it should, it’s supposed to be, it’s programmed to tell you what the source is. But as that sourcebook becomes ever more vast and the data points are ever more intertangled…

AI/ML Enters the Wild West

Yeah, I think that this is an important discussion. It’s one that I feel like we should have often as this technology is developing because this is the Wild West. There are no rules right now. We’re getting to see it play out live, and it’s gonna be interesting to see how it intersects and overlaps with the technology that we’re producing and, just generally in tech, how it plays out.

Kevin Bell: Yeah.

Joseph Drambarean: Yeah, this is so interesting.

Kevin Bell: Absolutely. No I haven’t seen a lot of finance applications of it yet. One of the funniest ones I saw was “Write me a poem involving the top 5 hockey players of all time” and it was miraculous to see Gretzky, Gordie Howe, and whoever else you want to put in there included in this awesome poem. Not quite the same as closing out your books or managing your cash but arguably a little bit more fun, maybe. 

Joseph Drambarean: Yeah, for sure. Well, thank you Kevin, this has been awesome. Thank you for joining us, this has been the Fintech Corner podcast. Find out more about Trovata at trovata.io and stick with us. We’ll be talking about more topics in tech, fintech, and the evolution of the finance space. We appreciate you Kevin for joining and we’ll see you next time.

Kevin Bell: Absolutely