Podcast Transcript: Artificial Intelligence and Asset Managers

Murray Bender: RBC Investor & Treasury Services is pleased to present insights on the future of asset and payment services across the globe. Today’s podcast features Mike Taylor, Head of Business Development at AlphaLayer, discussing artificial intelligence and its potential for driving innovation within the asset management business. Thanks for joining us, Mike.

Mike Taylor: Thanks very much, Murray. Pleased to be here.

Murray Bender: To start, Mike, can you tell us a little bit about your firm, AlphaLayer?

Mike Taylor: Sure. AlphaLayer is a Canadian applied artificial intelligence (AI) development firm that caters exclusively to asset managers. So for background, AlphaLayer is a joint venture between AltaML, which is an Edmonton-based applied AI studio, and the Alberta Investment Management Corporation, AIMCo, which is a $120 billion institutional asset manager for many of the province’s pension plans and endowments.

And the joint venture brings together AIMCo’s industry domain knowledge, and AltaML’s AI talent and expertise. And so we formed the joint venture in 2019, and since then we’ve deployed a host of applied AI-powered solutions for AIMCo as well as other clients—solutions that can be leveraged by asset owners, hedge funds, and traditional asset managers.

Murray Bender: So, Mike, what is artificial intelligence or AI? And what are some of the benefits to asset managers of utilizing AI within their firms?

Mike Taylor: Sure. Well, artificial intelligence, it’s quite a broad term for technology that mimics human intelligence. And we see AI in many of the devices and processes that we take for granted, such as self-driving cars, Amazon’s recommendation system, Apple Siri, anything Google searches, auto-completion functionality. The central benefit of AI is that it helps users make stronger decisions.

And this is definitely true for asset managers. A recent CFA Institute Research Foundation article on AI in investment management said that AI can—and I’ll quote here: “AI can produce better asset return and risk estimates, and solve portfolio optimization problems with complex constraints yielding portfolios with better out-of-sample performance compared with traditional approaches.”

So that gives a sense of where the value is. Maybe, if it’s helpful, I can drill down to a bit of detail about some of the components of AI and how that can bring benefit to asset managers.

There’s a couple of subsets of AI, machine learning and natural language processing, which deliver benefits to asset managers. Machine learning involves the creation of predictive models based on data from past examples. The machine learning differs from traditional software programming because it doesn’t compute based on a set of prescribed rules. It actually tips programming on its side by ingesting vast quantities of input and output data to define predictive rules and models using past examples to learn from. The strength of ML is that it offers opportunities for asset managers to create new data sets. And it can also support portfolio performance and the creation of stronger investment risk analytics.

So, natural language processing is a little different, also known as NLP. It interprets language, in text or oral form, and can support the conversion of unstructured data into usable data formats. So for investment managers or asset managers, NLP solutions can improve processing efficiency, volumes, and speed by electronically collecting data such as market data or commentary. And then it converts it into a usable format for sentiment analysis or even pushing to downstream systems for further analysis and processing.    That gives you a few examples of how AI can help asset managers.

Murray Bender: Just how prevalent is AI among asset managers?

Mike Taylor: Yeah. It’s quite prevalent. PwC recently reported that 84 percent of Canadian CEOs agree that AI will significantly change their business within the next five years. And in another PwC survey on productivity in financial services, they said that AI has passed robotic processing automation, or RPA, as the most widely used type of automation solution. So PwC is reporting that it’s quite prevalent.

But I do think it’s worth noting that there is a wide range of AI users amongst the asset managers, so there are some that are early adopters and others that haven’t yet taken the plunge. With regards to those that are early adopters, many of those firms are using AI for custom trading strategies, complex risk and analytics tools, and other AI-powered automation such as sentiment analysis or automating manual processes.

And one thing that we’re seeing recently for firms is that they are looking to work with AI to help with broader ESG challenges, such as factor validation and company analysis using clustering techniques. So that’s something that seems to be becoming more prevalent.

But that being said, some asset managers have not yet taken on AI within their operations or their investment management functions. And there’s a host of reasons for that. In some cases, they just haven’t identified use cases that are usable. And in other cases, they don’t know where to get started. So what we suggest to them is that the asset managers take the first step of performing a high level idea generation session, and then overlay that with a cost benefit analysis or an ROI assessment. And then they can build their business case for AI. And if there’s no business case there, at least they can tell their boards that they’ve done their homework, and that they’ve done the review of AI.

Murray Bender: What are some of the learnings that you’ve seen from these early adopters that you’ve just been mentioning?

Mike Taylor: Yeah. I think there’s probably three categories of learning. The first is around data. The second would be around managing expectations. And the third is around staffing. So maybe I’ll tackle each of those individually.

The first one around data. AI needs sufficient and appropriate data as input and to learn from. So, often applied AI projects can be stalled or halted because asset managers underestimate the data required. In 2021 MIT Technology Review reported that 13 percent of organizations excel at delivering on their data strategy. And they say that achieving data management by improving data quality and processing is critical to enabling growth-oriented efforts like those driven by ML. So I think that’s an indication that data’s really critical and in many cases asset managers may not be, yet, well positioned to support that. So that’s a learning, I think, for any asset manager.

The second is really managing expectations. I’ll go through a few of these a little bit faster. There’s four  of them that I have here. Many people believe that AI can solve every problem. And this is clearly a myth. So in the onset, I would suggest that any asset managers learn what AI is best suited for and what it’s not suited for before they go on the journey.

The second would be around defining business problems. So, many organizations chase the shiny, exciting AI solution before they even think about what is the problem that they’re trying to solve, almost getting ahead of themselves. So that’s definitely a learning. You want to know the problem before you start solving for it.Another learning would be focusing on quick wins. So, organizations like to have big wins; it’s very exciting. But in some cases it’s the small, quick wins that can build enthusiasm within an organization. Building that track record and then ultimately garnering the executive sponsorship is often a more prudent route to go for AI, especially when it’s not as well-known as some of the more traditional programming techniques.

And then the last expectations learning that I’d put out there is AI is an experimental learning process. The solutions are derived iteratively, and they may not end up as they were originally planned. So with any new technology, asset managers really need to manage their expectations and also gain the commitment of the senior sponsors to see it through.

So then the third learning is around staffing. I think the learning here is define your AI development staffing strategy early on. Attracting and retaining top talent, especially AI and machine learning talent, is super challenging and costly. So it’s not easy to find the right machine learning developers, data scientists, software developers, and project managers. So a classic build versus buy decision. There’s no right answer here. But if you choose to build your team internally, really recognize the challenges and the costs.

And then one add-on for staffing. Often the AI experts in the market are rarely deeply familiar with asset management. But this domain knowledge is critical for success for AI. So where a lot of organizations will put an AI team into an IT department, it’s an option for sure. But you want to make sure that the AI team are closely linked and aligned with the asset management subject manager experts. So it’s really important that they’re working collaboratively to define use cases and then ultimately the development. I hope this helps.

Murray Bender: Very helpful. Looking to the future, how do you expect that AI is going to evolve as we move forward?

Mike Taylor: One of the big movements right now is responsible AI. So in some ways it shadows aspects of the governance component of ESG. Responsible AI is the governance framework that documents how a specific organization is addressing the challenges around artificial intelligence, both from an ethical perspective as well as from a practical point of view. And the goal is to really mitigate the risk of something going wrong if there’s a problem with an AI initiative, and to help resolve the ambiguity for where the responsibility lies.

So for asset managers, this really plays out in two notable ways. One is around black boxes, and the other is around data bias. So for asset managers, the expectation is that the development and output of AI algorithms are well understood, and they operate properly. And responsible AI principles are driving the belief that it’s incumbent on the sponsoring business unit as well as the developer to make sure that the applied AI is explainable and trusted, so it doesn’t operate as a black box. And so that’s something that asset managers really need to be focused on is understanding what’s going on in the black box.

And then the second is around data bias. It’s probably more prevalent or understood in the retail space or wealth management where there’s demographic data. But concerning data bias, bias can be introduced into AI by the data that’s being used to train the machine learning model. And when training data is biased, decisions made by the programming are also biased. So this is something that not only asset managers, I think all industries really need to focus on. But I think that will become more a focus in asset management.

And then maybe one final thought on the future. A CFA Society webinar recently talked about the future of investment management. And it was super interesting to hear a senior leader from a large global asset manager comment on the need for ever-increasing amounts of data that requires the use of AI to perform many aspects of investment management. I think the underlying message there was that those asset managers that don’t have access to strong AI tools to make good decisions will really struggle to be top quartile performers, and potentially be left behind. So clearly a call to action for the industry. And I guess it’s an encouragement as well for asset managers to learn more about AI and consider how AI can help them make better-informed decisions.

Murray Bender: Thanks for sharing your insights on this emerging area of artificial intelligence, Mike. We really appreciate your time.

Mike Taylor: Thanks for the opportunity, Murray.

Murray Bender: Today’s podcast has been brought to you by RBC Investor & Treasury Services, and we hope you found it useful. For additional insights on the future of asset and payment services, including our previous podcasts, visit rbcits.com/insights. I’m Murray Bender. Thanks for listening.

This content is provided for general information and does not constitute financial, tax, legal, or accounting advice, and should not be relied upon in that regard. Neither RBC Investor & Treasury Services, nor its affiliates, accepts any liability for loss or damage arising from use of the information in this podcast.