AI Leadership Insights: Implementing AI in the Financial Industry

Amanda Razani: Hello, I’m Amanda Razani with Techstrong.ai, and I’m so excited to be here today with Suma Chander. She is the partner for PKF O’Connor Davies. How are you doing today?

Suma Chander: Very well, Amanda. I’m very happy to be here with you.

Amanda Razani: Wonderful. Can you share a little bit about the company you work for and what services do you provide?

Suma Chander: Absolutely. PKF O’Connor Davies is actually one of the top 25 accounting tax advisory firms. We’ve been around for a very, very long time, and we have a global presence. And I personally lead the PKF O’Connor Davies’ newly established digital transformation and advisory service practice, and we have multiple verticals. That includes artificial intelligence and data analytics. We offer outsourced CTO services. We also provide new software technology overviews. We offer compliance and tech reviews and a whole bunch of other initiatives around technology. So, I lead that practice for PKF O’Connor Davies.

Amanda Razani: Wonderful. Thank you for sharing that. That brings us into our topic of discussion today, which is implementation of AI. It does seem to be a technology that’s being really looked at carefully in the past year and a half or so, and all business leaders are trying to figure out how can AI help them, how can they utilize it in their business? What are you seeing in the industry and maybe specifically to the accounting and financial side first?

Suma Chander: I think there has been a lot of speculation, interest, but people are embracing and willing to adapt. What we are seeing is a lot of clients coming to us, asking for advice in terms of where do we start? How do we go about doing this? Where can we use artificial intelligence? What does it mean for us in terms of privacy? What are the various constraints that we should be thinking of? What are the ethics and rules around artificial intelligence? Where should we go to to start with? I see a lot of lack of education, but also, the internet is flooded with information. The news is flooded with artificial intelligence. There’s a lot of know-hows and don’t-knows on what to do and what not to do. So, a lot of the clients that currently we are already servicing are coming to us in terms of upgrades, how can we use AI within the current setting of what they’re already using, or people just want to now start embracing and say, “How can you help us to move from point A to point B?”

Within the accounting and financial sector, I can say that a lot of the investment banks, and also, the smaller boutique private equity companies, hedge fund companies and funds are embracing it with a little more zest because they’re already using artificial intelligence as a part of their tech stack that they already have, Microsoft or [inaudible] has come up with the Copilot. A lot of the AI features are already built. Azure has a lot of the tech AI built into it. So, people are already using it without knowing that they’re using it on one side. But a lot of the employees I see are starting to play with generative AI, especially ChatGPT and other tools, and they are experimenting. They’re trying to see if they can create models and simplify their work, read their contracts. So, yeah, we are seeing a lot of that.

Amanda Razani: Can you share some use cases, examples for us?

Suma Chander: Absolutely. Some of the potential work that right now, we are actually guiding clients to is in the marketing field. For example, in the marketing field, people are using extensive language. There’s a lot of extensive use of language, multi-use of multiple languages in terms of reaching out to audience, automatic emails starting to become redundant because these can be auto-produced. This area has particularly picked up on outreach in terms of outreach of either may that be a donor base, may that be a client base. A lot of that is getting automated, and a lot of the use cases that we are seeing is happening in that area.
There are also AI tools that now are able to integrate multiple documents, taking the highlights of multiple documents and summarizing it in the format that you want to see it. So, we are starting to see a lot of use of that where people are working, they’re reading through multiple contractual agreements, or say for example, they’ll have to summarize huge pages of data, and crunch them down into small paragraphs, and summarize the information to clients. People are using that in that regard.

People are using it extensively for data analytics, of course, in terms of how you model your data, how you train the data, in terms of specific outputs that you’re looking to have. People are starting to learn, to get sets of data from the outside and train it to give the specific output that they want. I see a lot of that happening within the portfolio management and data analytics space in the data mining space, and also, specific interpretation of how they want to use the data that they got. These are mostly where we are seeing. And of course, a lot of bots, a lot of chatbots. Pretty much everything is becoming automated in that space.

Amanda Razani: What are you seeing when it comes to the implementation and these digital transformation initiatives? What’s the biggest roadblock that some of these business leaders face during that stage, and how do they get past that?

Suma Chander: I think we are starting to advise, this is what we are saying. We advise companies to first look through at the function, the basic functions of their organization. We categorize them. Each group goes and comes up with use cases of what their daily work looks like. We take what is the most important things or what are the majority of their work that can potentially be automated using artificial intelligence. Then, companies are trying to steer themselves into maximizing the use of technology in these spaces in terms of efficiency or what AI can do. This has pretty much been the approach, but as they get… And also, defining what that point A is and what that end goal is, and defining the road path. That’s really the approach on how we guide clients on how to get there. We also present them the stack on the ways to approach.

There are tools that you can use that you can bring to the market that we can recommend depending on if it’s marketing, depending on if it’s a data analytics situation, depending on if it’s predictive analysis, depending on if it’s LLMs, just using generative AI for something. We tell them, you can use this for this, this for this. And then, we also train them and educate them on the challenges that they’re going to face on the way, may that be a new cybersecurity policy that they need to implement that they don’t have currently, may that be a crisis planning that they need to think about. What about the ethics around the AI? Is the data that they’re getting, because a lot of the models need to be trained, the data has to be accurate, it has to be comprehensive and governed properly. Otherwise, the outputs are not accurate. So, we help them in terms of making sure that the data sets are accurate. They get the full comprehensive data sets. These are things that companies need to think about, and they are. I think they’re already starting to adopt, and adapt, and train themselves, and there are implementation challenges that come up all the way because it’s changing so rapidly.

It’s not like a one-time implementation, and you can forget. There’s a lot of open source that’s being used out there. What does that mean to these companies if they’re using that open source as a part of their implementation? What kind of compliance challenges should they be thinking about in terms of regulation? Regulation is also changing. But also, AI is changing very rapidly. Now, they’ve come up with GPT-4 Enterprise, which is a completely closed model for an enterprise. So, a lot of the privacy issues that companies were having before, some of them have already been addressed through this model, so yes, I think both are changing at a very rapid pace.

Amanda Razani: I know in any industry, but definitely in the finance accounting space, I know security, you mentioned security is a big issue and the ethics. What should business leaders be thinking about when it comes to that standpoint and implementing AI?

Suma Chander: I think data is the key for everything. The kinds of data you get, how the data is governed, the right levels of encryption are now starting to be needed, for sure. A lot of dual authenticity in terms of applications, in terms of security logins of everything is now almost being mandated. I think rightfully so. A lot of cybersecurity and how it’s monitored is now… Actually, AI is actually also monitoring cyber. Artificial intelligence also has capabilities to monitor cyber problems as much as it’s creating potential for more attacks.

There’s a lot of new technology, especially what I mentioned right now. There’s new governance that is coming into place and certifications that are now starting to be adopted in the cyberspace especially. And data governance, and how data is encrypted, how data is saved within the databases, or data lakes, or that people are starting to now, the big data, what they call is also changing. Yeah, absolutely. And people have to think about it. People have to think about the backups, how data is transported, is it being safely saved in the database when the data is moving from one application to the other? Is it being encrypted, not encrypted? So, these are some of the things that people now have to think a lot more thoroughly than they didn’t think before.

Amanda Razani: Most definitely. As rapidly as this technology is advancing, where do you see AI in a year from now in the enterprise?

Suma Chander: I think AI now has recently come up with what they call multimodal AI, which means that now, it’s able to take text and image at once, and able to process it, and give you the output in a text. That is extreme processing speed. So, the move is happening more towards the hardware, and also, computing of how the processors can actually handle this with speed. The software itself, the research is breaking through every single problem that has surfaced within six months. According to an MIT professor, he stated that every single challenge has been broken and a solution found within six months of the problem arising. So, the speed is extremely fast. Now, they’re moving towards processing speed. The regular computer, and servers, and cloud computing, et cetera, may not be able to handle the speed, and also, of processing such large amounts of data.

So, now, they’re looking towards quantum computing and how that’s done, and bringing that into the chip technology, which is why there’s a lot of move towards NVIDIA. They’re doing super good because they are working on this technology. And I think the whole cloud computing and even including Azure, et cetera, they are now starting to move towards quantum computing, what we call quantum computing. I think the basic algorithms are around artificial intelligence, they’re getting broken very quickly, and the technology around in machine language and the algorithms are getting more and more advanced. And the more data that they’re going to learn how to train and make it better and better, the next level of evolution is now the hardware, and the processing, and the chips, and the ability to handle it at the speed that the software is now able to deliver already.

Amanda Razani: That will be interesting to see where quantum computing goes. That’s a whole other fascinating area.

Suma Chander: Yes, absolutely.

Amanda Razani: Well, thank you so much for coming on and sharing your insights with us today.

Suma Chander: Thank you so much, Amanda. It’s been a pleasure.

Amanda Razani: Look forward to speaking with you again soon.

Suma Chander: Thank you so much.