Better Together: Human-AI Partnership for Business Impact

Transcript

Katie Gaertner, Business Analytics Lecturer
How is AI being used in finance? Where do you see it playing out?

Erik
Yeah, another great question. I think AI, like a lot of fields, it’s touching all different areas of finance, and it’s sort of a horse race to see which areas it penetrates the most, where it makes the largest impact. So I think the most immediate or lowest hanging fruit are things like interactions with customers. Like when the customer has an inquiry about something that’s going on in banking. You can see chatbots will be very helpful here. These have been around for years, but now the sort of AI versions of them are souped up and really much more effective. This can lower the cost for financial institutions for handling some of the simpler inquiries from their customers. And, of course, those cost savings can get passed along to customers, so the customer can benefit as well. Some of us may not like interacting with the chatbot as much as with the real person, but if chatbots get better and better and they’re more effective, it’s really useful. So that’s sort of the lowest hanging fruit I would say. If you go one level up from that, you can think of something like a robo-advisor. Here you may have an individual who’s maybe investing for the first time and they’re looking for some financial advice, some basic sort of wealth management advice—how to manage your stocks, your bonds, your portfolio. Traditionally, a lot of this wealth management advice was only accessible to the highest income earners in society who could pay a financial advisor to help them with these decisions. But now we’re seeing this advent of robo-advisors. Companies like Vanguard or Fidelity or other large companies will offer these. Where you can get somewhat personalized, pretty useful advice at a very low cost if you’re willing to work with a robo-advisor, basically an AI-powered chatbot entity that’s really useful and can give good basic advice that’s somewhat tailored to your situation. So that’s really a democratization of a really useful financial service to a broader set of people that’s being powered by AI. That’s the second level. And then if you wanted to think what is sort of the cutting edge of where AI is starting to penetrate now, companies who are investing and making a lot of real-time trades and investment decisions, like you could think of a hedge fund. They’ve been incorporating some component of algorithmic trading for a while, but this is really intensifying, and you’re seeing AI being used to process in real time news and other sort of big data type information that comes in, and ultimately they use this in real-time investment decisions and trades. So that’s sort of the cutting edge of where AI is hitting in finance.

Katie
Interesting. So you see a lot of this predictive AI playing a role in finance?

Erik
Oh, definitely, yeah. Some of it will be maybe a bit more what we’d call machine learning than AI. But the definitions here are going to get blurred, and predictive AI and machine learning are going to mesh together, and eventually it’s a little bit black-boxy how all of this works, but ultimately really useful in predicting outcomes, or things like price movements in the short run that hedge funds would be incredibly interested in.

Transcript

Kaitlin
Yeah. So first I think I should be clear about what I think AI means. I think of AI as any algorithm that makes it so that computers can perform as well as, or maybe even better than, humans at a task.

Katie Gaertner, Business Analytics Lecturer
Great.

Kaitlin
A particular version of AI that has been transforming how we think about supply chains is computer vision. So think of this in a production setting. We have had a long history of moving from manual tasks to automated tasks, and computer vision is one more step on that path. Some factories have even gone what they call “dark,” so not relying on humans to perform their tasks, instead primarily relying on robots. Dark because you don’t need to have the lights on if it’s only run by robots. Similarly, in warehousing, we see a move from humans navigating the aisles, picking items off of the shelf to robots bringing the shelves to human pickers who then assemble your order at, say, an Amazon warehouse. And then transportation, right? A move from humans driving the trucks to self-driving vehicles. And all of this use of computer vision has a few, I think, important implications. One is about the layout of space, how do we use our limited resources, especially in a physical space kind of setting? Dark factories can have a smaller footprint, which means they can actually be closer to their end consumers. Same idea with warehouses that have these robots that are moving shelves around. Kind of a similar idea when you think about transportation. Self-driving cars, the ideal version, they actually communicate with each other. And so an individual vehicle can take up less road space, because you need to leave less of a buffer between them, which is particularly important when we think about dense urban settings.

Katie
Sure, yeah.

Kaitlin
So I think that this is going to have, I mean, it is already having, but it is going to continue to have a huge impact on how we think about supply chains.

Katie
Yeah, those are some really interesting examples. It’s really got my brain thinking about oh, wow, all of the things that are possible and things that might change. Maybe you can share with me some examples of how AI is changing the way companies compete and pushing companies to adapt their traditional business models to a new landscape.

Kaitlin
Absolutely. So my favorite example of this is the one that I study, which is a ride-hailing marketplace. So you think you want on-demand transportation from point A to point B. Twenty years ago, you had to rely on a taxi, and that taxi was operating without any AI or any kind of sophistication. You would stand on the curb, and you would stick your hand out, and you would try to hail them down. Maybe you would call a dispatcher who then would send out some message on a radio, but we’re talking very low-tech kind of dispatch. Then about 10 years ago, Uber enters the scene. Uber enters the scene, and they introduce a new service that is really based on this idea of new technology, right? We all have smartphones, and so Uber is able to keep track of the location of where I, the passenger, am, where all of the network of drivers are so that they can use an algorithm to perform these matches between folks who want a ride and drivers who want to provide a ride, which makes the service much more appealing in a lot of settings. Makes it so that I, as the passenger, don’t have to wait as long. The driver doesn’t have to spend a bunch of time wandering around empty, not getting paid. And there are a bunch of downstream implications from this that I think a lot of people appreciate about the service: the fact that you have automatic payment, and rating system, etc.

Transcript

Katie Gaertner, Business Analytics Lecturer
How is AI being used in accounting?

Ewelina
So this is really exciting. AI is essentially transforming this profession that has been around for so long that it has kind of been due for change, and it’s doing it in a lot of dimensions. So first of all, it’s automating a lot of these repetitive tasks that most people kind of found boring to begin with. And so it’s freeing up resources by automating these repetitive tasks so that accountants can focus more on the analysis and getting those deeper insights. It’s also improving the accuracy of the financial information that we’re using for a lot of decisions within our global economy. And so in terms of some kind of practical applications, in terms of the automation of data entry, there are tools that are now, I think it’s called optical character recognition. So instead of having people manually go through different documents and invoices, and things like that, it’s able to do all of that manual work and extract the really important information there. I know that the FBI uses really sophisticated machine algorithms to be able to look at the patterns in all of these different financial transactions and be able to notice some kind of suspicious patterns that might be indicative of fraud. It’s just so neat. You can be kind of a detective in some of these white collar crimes that have a really big impact on their victims. And it’s just a really neat opportunity there. And then, of course, in the area that I research, there’s a lot of companies that are using AI to improve their forecasting processes.

Katie
OK. And by forecasting, do you mean like forecasting revenues or forecasting demand? What are people forecasting?

Ewelina
Yeah, absolutely. Companies do a lot of different types of forecasting. They have forecasts externally, talking about what their earnings projections are going to be. They have a lot of internal forecasting, so that’s revenues but also earnings as well, to be able to help with resource allocation issues, to be able to budget, to be able to know where they should be putting human capital and what projects they should be investing in. And the forecasting really spans a great deal of context.

Katie
Very cool. Alright, so what are some of the most exciting, transformative AI developments that you’ve observed in accounting that you’re excited about?

Ewelina
I’m really excited about two things. The use of natural language processing is really taking off. And so information no longer has to be restricted or data analysis no longer has to be restricted to numerical information. So now we can use words and narratives and be able to analyze that and be able to triangulate that with the financials as well. And AI tools can analyze contracts, legal documents, emails, really any of this kind of unstructured data to be able to pair that with the financial information to improve decision-making. And then we’re also seeing a growing number of companies use generative AI. And we’ve got this evidence, they’re not really forthcoming with it, but we’ve got a lot of researchers that are kind of really digging in and finding evidence of this, that they’re using it to create these narrative reports. So their annual reports are now being, at least partly, or in a large part, driven by generative AI.

Transcript

Katie Gaertner, Business Analytics Lecturer
How do you see AI being used in management and human resources?

Jirs
It’s an interesting time to kind of think about this, because there’s a lot of experimentation. And I think that’s true across any jobs and fields where people are trying to figure out: Where is the potential of this? I think there are three broad areas. One is think about these routine tasks, these day-to-day things that just take up time. And I think a lot of companies are starting to think of how can we use AI to take some of that load off of our HR function so that they can focus on some of the strategic things a bit more? A good example would be in benefits: Giving every employee personalized benefit recommendations. Something changes in your family; what is the best configuration of health insurance? Retirement? All these kinds of things using AI rather than having people in HR necessarily spending a lot of time on that. Answering questions. Responding to emails. You have these frequently asked questions, can you have AI build that up? So it’s these day-to-day things that HR professionals do that I think they’re trying to include AI to have professionals spend less time on that. So I think that’s one area. Then the second is around strategic decision-making, particularly when it comes to data and data analytics. I think AI is lowering the barrier for a lot of companies, particularly smaller companies who don’t just have a massive amount of people who work in data analytics, and it’s lowering the barrier to more advanced analytics. Thinking about turnover, for example, trying to predict who’s actually leaving? Why are people leaving? And doing these more advanced analytics on that that usually only was reserved for companies either who had a lot of people to do these, who had these skills, or who could hire McKinsey to come in and actually do that. But now these mid-sized, smaller companies get access to that because of AI. You can think about managing performance. We need to increase or try to optimize performance in our workplace. Well, can we develop personalized training and development programs for every employee depending on what they’re doing, what they’re working on, what their tasks are at that moment? That we can think about that. And then I think the third area, which is a little bit more controversial, is what I call managing people through AI. So think Uber for example. Who’s your manager? Well, it’s the AI. It’s a computer. It’s the app. And so I think right now that’s a little bit more reserved for these kinds of companies like Uber, you know, those kind of things, managing, getting managed through the app, through the computer. But you could imagine that maybe companies are going to start experimenting with that as well: How can we manage our employees using these AI functions? Which is a little bit more controversial, I think, but it’s another area that I see kind of AI starting to be used in HR.

Transcript

Katie Gaertner, Business Analytics Lecturer
What are some of the most exciting or transformative AI developments that you see coming up in marketing that you’re really excited about?

Neeraj
I think the answer is not gonna surprise you. And the answer is generative AI. So all the examples I gave you I would call traditional AI, and I would say, just making up a number, that’s like 70-80% of AI today—well, that’s called the traditional AI, all the examples that I gave you. Where the real sort of excitement today is is in the world of generative AI, and in particular, large language models. And what’s happened in that space over the last two years with GPT becoming so mainstream is that you really are in a brand new world. Let me give you a couple of examples. Any action that a company takes is typically based upon insights they glean from consumers. So what are consumers or their needs? What price point’s appropriate? That requires a whole sort of slew of marketing research tools. So what we show, and this is one paper that recently got published, is this whole pipeline of the research process—from problem recognition to which research approach do I use, who to sample, what data to use, how to analyze that data—this whole pipeline, every step along the way, we show in this paper, LLMs can really be your friend, your assistant. And then they just do a phenomenal job of, for example, picking who to talk to, what kind of questions to ask, stuff that was like really monotonous, mundane, repetitive tasks that people were writing surveys, for example, or writing a discussion guide for for qualitative research. All that stuff is now very heavily assisted by these generative AI LLMs. And the two places where I see big traction that’s happening. The first one that won’t surprise you, because you’re the analytics person, is in the analytics space. We have companies that I’m now working with, they’re finding that they’re super able to streamline analytics so much that that stuff that used to take a day to do can now be done in an hour. It’s like these order of magnitude gains are super impressive. The other one that may be a little less obvious to our students is this notion of synthetic respondents. So in the traditional world I would go find people and talk to them and get their data and analyze it, qualitative or quantitative, and then glean insights from that. But what if we were in a world where I could create a persona. So I could create a persona for Katie that I look at your background, your education, the fact that you’re a professor, you teach analytics. I could make up this persona and then ask the LLM to ask your evaluation for a variety of products. So we do that in that paper, and there’s this whole sort of new world of vendors out there that are playing in that area. So that’s very disruptive.