Episode 63: Morningstar’s Mike Allen on AI

Episode 63

Hello, and welcome to Revamping Retirement, a podcast brought to you by CAPTRUST, where we explore the opportunities and challenges facing today’s retirement plan sponsors and fiduciaries. Our hosts lead the employer-sponsored retirement plan practice at CAPTRUST, one of the largest registered investment advisors in the U.S., and a thought leader in the retirement plan advisory and consulting space. We hope you enjoy Revamping Retirement.

Matt Patrick: Hello, and welcome to another episode of Revamping Retirement. I am Matt Patrick, and I am joined once again by Dawn McPherson. Dawn, how are you doing today?

Dawn McPherson: Matt, it’s great to be back with you again. We’ve become quite the duo for these podcasts.

I’ve heard whispers of the Dream Team around the office; I don’t want to put that label on us, but others are saying it.

Matt Patrick: I’m excited for our conversation today. We are speaking with Mike Allen, who is the head of technology for Workplace Solutions at Morningstar. And our focus is going to be on artificial intelligence and how it might shape the retirement plan landscape moving forward.

But before we welcome Mike to the show, Dawn, I want to get your thoughts about and your experience with AI.

Are you using it? If so, how are you using it? It’s so in the news now. I want to get a sense for day to day, what that looks like for you.

Dawn McPherson: So, I would say I have lightly explored AI, and it’s been in a very nonthreatening way because I haven’t been dependent on whatever results it produced. I was almost testing it out. The first few things I used it for were things like writing emails. I wouldn’t spend a lot of time on the language, and just write a message and then either ask to clean it up or make it more concise, something like that.

Matt Patrick: I feel like you sold yourself short when you said you didn’t want to be graded on it. I thought you were going to say ah, you know, I’ve dabbled once. We did three pretty productive work things. Now I feel like I’m going to look bad with it.

Dawn McPherson: Don’t embarrass us, Matt.

Matt Patrick: I feel like for the first year I experimented with it [it was the] parlor tricks version of I can get it to write something funny or I can get an image of a dog on Jupiter, in the style of this famous painter.

I feel like I did a lot of those, and it was fun. We did trivia one night and the group couldn’t choose a name. Asked ChatGPT. QuizTeam Aguilera is the name that it came up with and it was wonderful. So it was really good at that. But I feel like it’s only been in the last few months that I’ve been starting to make that pivot that you talked about, of trying to look for more productive ways to use it.

And it’s almost like for a while I felt intimidated by how many people are saying it’s such an efficiency improvement and I’m doubling my output already. I’m just not using it right to achieve the outcome. It was almost safer to be in the fun elements of it, but I’m starting to see some of the day-to-day benefits, like you’re saying, like process efficiency improvements or grammar checks or rewording certain things. Or, if you get stuck, just having the words on the page to get you started has been a big help.

So I’m starting to envision and starting to buy into all the benefits it could have over the long term.

All right. Transitioning now, we have Mike Allen on with us. Mike, welcome to the show.

Mike Allen: Hi, Matt. Hi, Dawn. Glad to be here.

Matt Patrick: So we gave a very brief intro before you hopped on, just read your title. So maybe could you explain to us what that means, what you do at Morningstar and why you might be the person we’re reaching out to as it relates to artificial intelligence.

Mike Allen: Yeah, sure. I head up the products and technology organization in the Morningstar Retirement P&L. So, we’re focused on managed accounts products, user experiences for participants, our AMA product, and building software that hopefully gets participants to a successful retirement outcome.

Dawn McPherson: Before you joined us, Mike, Matt and I did a little pre-show where we talked about our level of familiarity with AI and how we’ve been using it, if at all. What we’re curious about is how did you get interested in AI? And then maybe a little bit more about how you’re currently using it.

Mike Allen: Sure. Yeah. I’ve been interested in AI for a long time. I’ve been in technology my whole career. So, I’ve been in cybersecurity before, and I spent a lot of time in cybersecurity, and we were using some AI tools and cybersecurity to kind of filter out noise and try to figure out who’s attacking us and what might be a false positive versus not.

So it was interesting stuff back then. But it was kind of just, we’re experimenting with it as a hobby. I remember we were at a conference, and Matt, I think you were there. We were talking a little bit about AI, and this is before ChatGPT came out. And, I think it’s like, Hey, can we do something cool in the AI space that would increase participant motivation and engagement and make things more relatable, make things more human.

And then a few months later, ChatGPT was announced and the whole world exploded. And then we kind of just kept going, and I think everybody started talking about AI then.

I’m using it more and more since ChatGPT came out. Brainstorming has been a really powerful thing. So being able to just bounce ideas off of ChatGPT or Google’s Gemini or other AI products. I’ve been using it for brainstorming, just kind of saying, Hey, what do you think about this?

And it gives me some answers. And then you just continue the conversation. It’s been, I thought, a pretty effective way to do it. And we’ve been using it kind of collaboratively too, just to kind of ignite ourselves with some ideas and get some conversation started.

You want me to keep going? I have a few other examples.

Matt Patrick: Sure. Yeah.

Dawn McPherson: Please share some other examples.

Mike Allen: All right. So, one other one that was really interesting, I bought my son an Oculus Rift for Christmas, so it’s pretty cool. He’s like, Oh, this is cool. But he’s like, How much does it cost?

So I told him how much it costs. He’s like, I actually want you to return it. And I want to invest the money you gave me, which … I was floored. My son’s 10. I guess some kids at school were getting him into investing. So, I’m like, okay, this is cool. Then he’s like, I want to return the money. I want to invest in the stock market. And he knows nothing about the stock market except what he’s heard at school.

So I’m like, okay, what do you know about the stock market? Didn’t know much.

He’s really into these Big Nate books and he’ll power through them. He loves them and just reads through a few a week, I feel like he gets through. So, I’m like, how can I teach him about the stock market?

I got him a book. He didn’t read it. It wasn’t as interesting as Big Nate. So I went to the AI stuff and I said, Hey, what if we can teach him about some of the stock market concepts and investing in the format of Big Nate? I used some of the AI tools to create a book that sounded like Big Nat, but was around the stock market theme.

So, he was able to relate, and he loved it and he’s able to kind of read it. [It] had similar tones and humor as the Big Nate books, but it was all teaching him about the stock market. And then I used some images through the image generators to create images that were relevant to what I was teaching him in the book. And it made it more relatable to him and fun to do.

Matt Patrick: That’s amazing.

I feel like I was talking at the front end of the show about how I’ve played around with the tools kind of in that way of like write something like this and this style or whatever. That’s an actual practical way of doing that. And you got a good response for it. That’s awesome.

Mike Allen: Yeah, I thought it was really cool.

They did a really good job, actually. It was surprisingly good.

Matt Patrick: Long story short, he’s running this very successful portfolio now. And it’s all going great. You can retire early.

Mike Allen: I don’t know if we’re there yet. I gave him a bunch of virtual currency to experiment with, but we’ll see how it goes.

Dawn McPherson: Maybe in the follow-up session, we can see how his investments have done.

Mike Allen: That’s right. Yes, we could do that.

Matt Patrick: Well, that’s awesome. It’s great to hear that you’ve been seeing a lot of really good impacts in it in your personal life.

I guess pivoting more from a work standpoint and transitioning more to the retirement industry itself, I think we’re all hearing AI, it’s going to change everything. It’s going to change the way we work. It’s going to change how we interact with all of our programs.

What do you think specifically for retirements? What do you think will have the biggest impact there from an artificial intelligence standpoint?

Mike Allen: Yeah, retirement, there’s definitely a lot of use cases I can think of around the retirement space. AI in general is going to change a lot of different industries. And I’ve been thinking about this question, just how it’s going to impact my own job, the products we’re releasing, and the industry as a whole.

And I think kind of building on the Big Nate example, I look back at what happened in the last like year and a half, two years. There was a lot of AI research and it was really cool. There’s a lot of cool stuff out there, but it was kind of not mainstream, if you will. Then ChatGPT came out and all of a sudden overnight it went mainstream. I think ChatGPT did such a great job at executing a product and that magic moment that it made it relatable to the average person Like, wow, this is really cool. I can see how this can impact things in the future in a way that they weren’t able to do that before, even though the research was out there for a while before.

So when I kind of look at the retirement space, I think about how we can do that maybe for the participant experience. I think one thing that’s really powerful about going to an advisor, for example, for people that can go to an advisor, I think you get that personalized, one-on-one relationship.

They understand your goals and get to know a little bit about you and kind of build that overall relationship and say, okay, what’s important to you? How much money do you have? What are your goals? And then they build a customized plan that you kind of work through together over time. And that human element is really powerful to keep you on track and keep you accountable. Ultimately, hopefully, it’s that goal you set.

And what I’m most excited about with the AI technology that’s coming out is the ability to create a more natural humanlike experience similar to what you get from maybe an advisor, but scale it to people that aren’t eligible to even talk to an advisor, which wouldn’t be an option otherwise for them. Get a very hyper-personalized experience that does a lot of the same things in a form that they understand and can naturally interface with, like you can with an advisor.

And I think you see what’s happening with ChatGPT, the multimodal experiences, the voice communication, natural language processing that’s happening. It allows you to create that experience, but at a scale that you couldn’t do before.

Matt Patrick: You said multimodal? Easy for me to say. Could you explain what that is?

Mike Allen: Yeah. I mean, the ability for AI to process using images or text or a video. Now with the release of Sora, you can create dynamic video. So the ability to just be able to interact with multi-formatted things like video and text.

Matt Patrick: Okay. You could input different formats of information and it can process it.

Mike Allen: That’s right. Yeah.

Dawn McPherson: So, you’ve said a lot related to participants: personalized tools, resources, a more human feeling, natural interaction. I might want to come back to some real-life application because you mentioned that too, but for right now, let’s keep on the theme of impact to the retirement industry. I think the focus on the participant has shifted for plan sponsors on what can we provide to participants to help them along? All kinds of tools and resources around advice and just encouraging better outcomes. What do you see in the way of maybe administrative or operational lift for recordkeepers or plan sponsors, as it relates to AI’s impact?

Mike Allen: I think from just a recordkeeper perspective, AI is going to cause us to rethink a lot of our existing processes. And I think there’s path number one, do you understand the technology? Path number two is do you actually know what it can accomplish?

And then it’s kind of like revisiting almost from a first principle standpoint. Like, how did we design this originally using the legacy tech stack? And what can we do with this AI tech stack now that we understand it, to make our process very different from what it was today and hopefully way more efficient.

So, if I use the recordkeeper example, I think there’s a lot of things that recordkeepers could leverage AI for to make process efficiencies. If we’re going to interact with a third-party system, can we revisit how we’re processing the data and use AI to make the processing of that data way more efficient?

Can I build an agent that goes out and connects to third-party systems that do things that I couldn’t do before?

There’s a magnitude of different things I think that the recordkeeper space would probably use. I think just being able to modernize all of our systems to take advantage of being able to securely exchange data and interface with these different agents more efficiently, so we could build the next super app or the killer app that allows us to take advantage of more automation.

The things AI will enable us to ultimately do in the personalization space is something that we have to keep iterating toward. From a sponsor perspective, I think just looking at the benefits, the plan. Or virtual advisor as a service to all your plan participants through the use of some of that personalization. Can you get better engagements, usage out of the product so you can see the scores go up and how people are interacting with things, the bars moving in terms of the outcomes they’ve achieved. I think there’s just a lot of potential there to improve from what we have today.

Matt Patrick: That’s great. I think that’s perfect. And I think it’s interesting to highlight [that] most people hear AI and they’re used to ChatGPT, so I think they’re used to that front end interaction. There’s a lot of people selling it on the idea that yeah, you’ll get advice from AI or that’ll be the agent that you’re speaking to.

I guess I’m fascinated by the first wave of improvements will be on some of those back end processes. Maybe things you don’t see, but that just manifest themselves, like, Hey, the service feels like it’s improved in this element of the industry, or I’m getting responses quicker, this just seems to process in a cleaner format, or I have more choice in terms of how I’m getting my data from the recordkeeper or the inputs I can give them. I’m fascinated to see what the progression looks like there, where I feel like people are like, Oh, they’re gonna make me talk to a robot and maybe I don’t want to. And that’s what you think of first, but maybe there’s a bunch of interactions or ways that you’re going to experience the AI influence and you don’t even realize that that’s what’s happening behind the scenes.

Mike Allen: Yeah, yeah. And I think just building on that, I think AI, it’s early days. There’s lots that we need to do to figure out how to make that client-facing experience better; to really work well and be trustworthy. Know that the data you’re getting back is something you can trust and rely upon.

So, I think there’s a lot of things that need to happen before we just roll us out of that at mass scale to make sure that we’re comfortable with all that, controls. At Morningstar, we have an AI governance committee and we take those things very seriously where we set up some guiding principles around what’s important to us and how are we transparent with how we process this.

We have a lot of discussion on biases and how do we want to handle that. So, I think a lot of that foundational work needs to be sorted out just before you roll out. And compliance is another key area that you have to look at.

Like, are we okay using this software in advice settings or providing advice? I don’t think we’re there yet, but eventually, maybe we will be there once we build all that foundational work. And I think the opportunity now is how do you build, make our internal process more efficient?

How do you ingest the data sets that we have that we know we can trust? Bring those internally, help people work as a copilot to kind of help them better make decisions, get access to information faster. And as we mature these things over time, I think we’ll see those participant experiences enhanced too.

Matt Patrick: Can we pick up the governance piece that you mentioned in there? Cause I know big debate across the industry is more on those biases, the governance, who gets to design the rules. Whose bias is there in the machine? Should it be Google’s? Is it open AI? Should it be the government? Who gets to set the rules there? I guess as we think about the impact of this in the retirement plan space, who gets to decide that? Is it Morningstar? Is it the plan providers? Is that going to be something that trickles down to plan sponsors in terms of the tools they’re picking and what biases are in there? And does it align with their goals?

What are your thoughts on that?

Mike Allen: Yeah, it’s a tough topic. I think we’re at the early days to try to figure that out and come up with something that works. I mean, I’m sure you’ve seen some of the headlines recently where that can go very badly, when you introduce too much bias and things like that.

So I think there’s always been a certain amount of bias present or whatever it would be, even before AI. I think you should look at when anyone’s writing something, there’s some perceived bias. I just don’t know when you’re kind of training a system that a lot of people are using, it becomes really a much more important topic to get right. And I have thought a lot about this and I don’t have an easy answer to say whose bias is correct.

It’s a hard question, Matt.

Matt Patrick: Yeah; we get to ask the questions of you and I don’t have an answer for it.

Mike Allen: Let me flip that around. Do you have an answer?

Matt Patrick: No one in any part of the AI industry has an answer for it. So why not you, Mike? How about you?

Mike Allen: I do not have a great answer. I know we’ve spent a lot of time discussing it internally. We’re trying to come up with a framework ourselves, but I think this is something the industry is struggling with.

Dawn McPherson: Matt, did you have a follow-up? I was going to let Mike off the hook and pivot back to another topic we already touched on.

Matt Patrick: It’s interesting, because it feels like the retirement industry will follow some of what we’re seeing elsewhere.

Morningstar will be building something. You’ve got internal governance. I’m sure all the other players will have some version of internal governance. But then I do think plan sponsors are going to have to grapple with how does each of these tools work? And what advice am I going to get here? And does that align with my view?

And, essentially, you’re going to be matching yourself up from a view’s perspective to align there. And then who knows if regulators come in and stipulate certain things. I think that’s looming over AI right now.

And I’d imagine as we get more comfortable as an industry with leveraging some of those tools and they get introduced, it feels like we’ll go through the exact same process. And we’ll just have to wait and see how that plays out and what it looks like for plain sponsors.

All right, Dawn. I’m done now.

Dawn McPherson: No, that was good. And it does seem, to that point, like it’s only a matter of time before there would be some type of establishment of best practices. You said it much more eloquently than I attempted to.

Mike Allen: That was very well put, Matt.

Dawn McPherson: I was just going to circle back. We talked a little bit about the efficiencies with plan providers and asset managers, potentially some of the efficiencies AI can provide. We talked about it benefiting participants. I think we can more broadly offer some of these tools and resources to participants, and I think Matt touched on the fact that service levels might improve, and we’ve struggled with those in the past few years across the industry.

Are there any other ways that we haven’t called out that you see AI and the efficiencies with providers and asset managers improving the sponsor and participant experience?

Mike Allen: I’ve been thinking a lot about how before it even gets to the product, even how we build software from the start. We have the software engineers that are creating this. And traditionally [this causes a] bottleneck in trying to release a lot of software. We just have a shortage of capable engineers that we can hire in short order to do this creation of the software that we need to make things more efficient.

And looking at some of the tools last week, a week or two ago, Devin, the first fully AI engineer, was announced. And I thought that was a really interesting announcement.

GitHub Copilot and a lot of different tools help make developer workflows way more efficient than they’ve ever been in the past. If you’re a junior developer, it’s never been a better time, in my opinion, to just get into software engineering and kind of help.

It helps you learn a lot faster than you would have, otherwise. So we’re starting to see almost a 10x developer model and a lot of just seeing even more junior engineers because they’re able to access these tools to help them kind of create software way more efficiently.

So, just starting there is an efficiency play. Just being able to create and automate a lot more for both the sponsors, the RAs, the recordkeepers. I think that alone is going to help level everybody up and create a lot more tools that we just didn’t have access to previously.

I’m really excited about that. And then just building on that, I think it’s like the copilot element. It’s almost like the copilot can have superpowers where you have this knowledge that an RA has, but also, you can ingest your entire knowledge base that you guys at CAPTRUST have, or we at Morningstar have. They have really cool things coming out where it can even listen to conversations and then start overlaying this knowledge right over your screen as someone’s talking, that might be relevant to the conversation.

So when you have things like that happen, all of a sudden things get a lot more productive because you don’t have to say, Oh, let me take that away and go research it. It’s right in front of you as you’re talking about it. And that kind of gets you the right context, the right time you need it, and just makes everything way more effective.

Dawn McPherson: Are you saying that someone could just listen to me speak and then add knowledge? Like if I have to speak on a webinar, they’ll just create that script for me? Wouldn’t that be amazing?

Mike Allen: Awesome, right?

Matt Patrick: I feel like the few times that I’ve stumbled over my questions, maybe it’d pop into, “I think you were trying to ask this, and these are the words that you need right now.”

Dawn McPherson: Oh my gosh, that would be handy.

Matt Patrick: That’d be great.

You had mentioned up at the top that you were working some avatars and generative video, that type of stuff.

Are you still a believer in that? Or what’s the current state of thinking there?

Mike Allen: So it’s interesting, the avatar was something that I was really bullish on at the beginning. I’d actually joined some AI focus groups where they’re talking a lot about avatars recently. And funny enough, it is the human avatars–and this is a personal opinion; I personally liked them. But I think, in user testing, some of them just weren’t good enough to convince someone it was a real human.

So, they found it a little creepy. Instead of using a real human, the suggestion was to use a cartoon avatar instead. So don’t try to make it be really human, but use something like a cartoon avatar instead to explain something. So in that respect, they’ll trust it a little bit more because they’re not trying to be human, when they’re not.

Matt Patrick: Did that work, switching into a nonhuman avatar?

Mike Allen: We haven’t done any user testing yet, but it’s something on my to-do list. We’re going to try out a little bit more, but I think explaining concepts using some type of avatar is useful to some people. And I think some people like to consume information that way. And others prefer the text, and others the traditional web format.

I think that’s the nice thing about the AI side. It lets you kind of adapt almost what the person would like to consume the format in.

Matt Patrick: I can envision the customization available in terms of what cartoon character would you like to give you financial advice. And so I’ll have SpongeBob give me a rundown of what I need. I can envision it right now.

Dawn McPherson: Have you been on calls, though? We have the Teams app. And have you been on calls where people substitute their face because maybe they’re working from home that day or something with an avatar? I don’t love it. It’s weird to me, I think it’s because I know the person’s there. If it was in place of a person, maybe I would like it better. I don’t know. It would be interesting to see what test groups feel about it.

Matt Patrick: If everybody was using an avatar, would it be better?

Dawn McPherson: I don’t know.

Mike Allen: Or the Apple Vision Pro model where they take a picture of your face and create an avatar that looks like you. Is that better? I don’t know.

Matt Patrick: Yeah. Maybe I could sand down the edges, that’d be great. That’s the best version.

I know something we had talked about in prep, there’s some posts out there around pulling together all these different elements and the idea of building an autopilot for your money.

We know default features work in retirement plans. It’s been one of the most successful things in terms of getting money into plans, getting people better prepared for retirement. But this is maybe an extension of that. Could you walk through some of what that money-on-autopilot idea is and what that might look like?

Mike Allen: I think you’re talking about a blog post on a16Z, right?

I mean, it’s interesting. You look at the retirement industry and the whole concept of defaulting in, and I think at the end of the day, when people think about finances, a lot of people get turned off.

They don’t want to spend any time on it, but they know it’s important. They need to do something. So how can you make that a lot easier for someone? And I think with the advent of some of this technology, being able to make it relatable so they understand why you’re doing something, explain it in clear terms that they are able to consume maybe, in a format that with their personality they prefer would be great.

And then once you’ve kind of explained it and they opt in to it, or they agree with it, how can you automate it and just take that off the table. I think that there’s a lot of interesting things that are going to be possible, I think with some of the agents that AI has, where you can just go out and say, Hey, I want to have a refi agent where it goes out and looks for the best possible loan rate that might help me save some money. And it’s up to the agent to kind of go out and figure that out and then recommend it back to you. Or maybe you just even take action eventually, if you’re comfortable with that.

The ultimate goal there is can it do the research for you and provide you with information that you understand and you can take action on easily. And I think that will ultimately make people do the right things with their money and take some of that traditional burden that’s been there, the inertia factor, away.

Dawn McPherson: Matt may have more specific comments about this, but we talked a little bit about it. I think people hear about AI, just like anything else when it’s unfamiliar. It’s a new tool, system, technology, whatever it is. You hear about it and it’s daunting and you keep hearing that, oh, AI is going to replace people and jobs and things like that.

But what I’m hearing throughout the conversation today is really that it’s a lift and a support and it helps create some scale and efficiency and offload some of the operational. I’m sure there are a lot of ways we’re not even thinking about it yet, but to me it seems more like an assist. Especially even in the way you just described it, with helping to deliver information in an easily consumable way for us to make just these everyday decisions a little more efficient and practical.

Mike Allen: Yeah. And that’s certainly the way I’m looking at things now. I don’t think it’s ready to replace a lot of different things cause it’s not reliable enough in a lot of ways. It hallucinates and there’s guardrails you want to make sure you have in place.

A lot of the work we’re doing now is paving that foundation so you can get more comfortable with things. I do think as a superpower or copilot, it works fantastic as long as you have that human element knowing that it could hallucinate.

You have the subject matter expert review it, make sure it’s factually correct. And you feel comfortable providing that advice. I think that’s a great use case for that now, and things we’ve been playing around with are mostly around that use case. And I think it helps make a lot of people way more efficient than they have been because of that.

Matt Patrick: Will you define AI hallucination?

Mike Allen: Yeah. AI hallucination is funny. Like you ask AI something and it comes up with an answer that is incorrect, factually incorrect, but it does so very confidently. So you believe it to be correct. It’s my experience and I’ve definitely had it hallucinate on me a bunch of times.

I think this is why it’s important to fact check everything you’re getting from the AI, at least now, until we’re able to perfect that.

Matt Patrick: I’m with you. It is delivered in a way where you’re like, well, that’s just how it happens. Those are the facts right there.

Mike Allen: Very confident about the wrong answer. Yes.

Matt Patrick: It works.

Dawn McPherson: Isn’t that how we all should be? Confident in what we’re delivering?

Okay, Mike, sticking with this market type or financial backdrop here, let’s play a quick little game called Buy Sell.

Mike Allen: Okay.

Dawn McPherson: If you agree with a statement that Matt says, you can say buy. We might ask you for clarification.

If you disagree, you say sell.

Matt Patrick: Perfect. All right.

Dawn McPherson: All right, take it away.

Matt Patrick: Our first statement, AI will increase privacy and data security risks.

Mike Allen: Will increase privacy and data security risk?. I think that is true to a certain extent. I’ll use an example. If you use AI to craft personalized phishing emails and like that, it’s very good at doing that. And you can personalize it and maybe trick someone into doing something.

So I think there’s definitely heightened risk there.

Maybe you can combat that with some other AI systems that are fighting back. So again, some of those malicious use cases.

Matt Patrick: You’re trying to, he’s trying to middle it down. I don’t know.

Dawn McPherson: I was going to say, is that a hold? He doesn’t have advice, so he’s just a hold? I don’t know.

Mike Allen: Just so I’m clear, when I buy it, am I saying, yes, I agree with it?

Is that right?

Matt Patrick: Yeah.

Mike Allen: Yes. I buy it.

Matt Patrick: There we go. Well, I think those are important because maybe that scares people away. I think it’s just a fact. It’s going to change the data security environment. And I think we know that. And you have to adjust because whether you’re intentionally doing it or not, you’re going to interact with more AI tools.

You’re going to get these email phishing attacks and they’re going to be more convincing than ever. So, even though you’re buying it, I think that’s a fair approach. All right.

Statement number two, people will trust information coming from artificial intelligence as much as they trust the information coming from a person.

Mike Allen: I’m going to sell that at the moment. I think people understand it hallucinates, the people who use it today. And I think they understand they probably need it. If it’s important, they’re going to do their research.

Matt Patrick: Makes sense. Makes me think of like early days of Wikipedia too, where they’re like, well, you can’t trust it. I don’t know. I still felt like I trusted it more often than I didn’t. So, it’s good to be aware of the hallucination piece. All right.

Statement three, most content consumed online will be AI-generated.

Mike Allen: I sure hope not. I’m going to buy that. I think there’s going to be a lot of AI-produced content in the near term. Yes.

Matt Patrick: Seems like it’s trending that way.

All right. AI image and video generation will be more impactful than some of the language models.

Mike Allen: I’m also going to buy that. I’m just seeing some of the Sora stuff coming out of the Open AI stuff, and it’s looking incredible. So I think there’s a lot of potential.

Matt Patrick: Yeah, it feels dangerous, though. Some of those videos, they’re saying, what, it’s like Pixar-quality videos that are generated just from your prompt in a couple of seconds.

Mike Allen: It’s amazing. Yeah.

The samples look amazing. If that actually ends up being the real production system, it’s going to be awesome.

Matt Patrick: All right. Last one. And this ties back to Dawn’s question.

Think about it taking things away or reducing things. AI is going to reduce the number of jobs available.

Mike Allen: I think it’s going to change jobs. I’m not sure it’s going to reduce the need for jobs. I think if you’re doing something that you might be used to today, I think AI is going to make it much more efficient and maybe push you into something that you weren’t expecting or a higher-value task.

So I don’t think it’s going to eliminate jobs. I think it’s just going to change jobs. When we had farming, it created a lot more machinery, but there’s still farms and still jobs there. It’s just a different job that people are doing. I think it’s the same thing with AI.

Matt Patrick: All right. Beautiful. All right. Our first round of Buy Sell, we had

Buy: AI will increase data security risks.
Sell: That people will trust information coming from AI as much as people.
Buy: That most content online will be AI-generated.
Buy: That AI image generation will be more impactful than language models.
And sell: The idea that it will reduce jobs. It’s just going to change jobs. So that makes sense.

Dawn McPherson: Very well done. Okay, so I’m going to ask you one more question that might feel a little doom and gloom, and then we’ll end on a high note. But worst case scenario, where does this just go off the rails?

Mike Allen: I was listening to a podcast about this exact thing yesterday. And I also recently watched the Terminator with my son.

Dawn McPherson: Very appropriate.

Matt Patrick: What can we learn about stock market investing from Terminator? It feels like that’s what’s coming.

Mike Allen: That’s a good question. I can ask AI.

Mike Allen: I think there’s a lot of cases where AI can go good and there’s definitely some I can think of where it can go really bad. I’m hoping we’re going to use AI and it’s going to kind of go in the good camp.

I think this is why it’s really important for us to think through the principles and transparency and the guardrails that we want as we go through and create these technologies. And I know a lot of companies that are leading in the space are paying attention to this.

And I think it’s important that we prioritize that sometimes over speed. A lot of the CEOs that I hear talk in this space are definitely prioritizing this now, or at least there’s a lot of conversations about it happening.

I’m hopeful that as an industry we’ll prioritize the right things, put the right guardrails in place, but I think there’s definitely a non-zero chance that something bad can happen. I think we’ll definitely have some hiccups along the way. But I’m hopeful humanity will adjust and make the decisions necessary to fix it.

So I’m positive on it. But I definitely think there’s a risk, a non-zero risk.

Matt Patrick: Great. Well, we really appreciate you taking the time to walk us through your experience with AI and all the exciting items to keep an eye out for on the horizon.

I guess we’ll close with a more personal lens. This is a question that we ask everybody that comes on the podcast.

We want to know what retirement looks like for you, Mike.

Mike Allen: What retirement looks like for me. Yeah, it’s a good question. I think I’m excited about all this AI tech and the augmented reality stuff. I think retirement is going to be awesome. I’m going to be playing with tech toys the entire time. But more serious though. I have two kids, so I’d like to be close to them.

I’d like to be in a warm climate. My wife likes cold, so it’s going to be a little bit of an arm wrestle there. I’ve grown up in cold. I want to move somewhere a little warm, hopefully close to the kids.

I’m a real tech guy at heart, so just play around with this AI tech. I think we’re going to be a lot better placed by the time I’m retired. I think if I want to go somewhere different, I can just strap on a headset and go visit wherever I want in the world, I can see those days coming and I’m excited about our robotic chefs being present in my retirement. Imagine where you could just go to a restaurant you like, and you can just license a recipe from them and your robotic chef will just make it in your kitchen. I’m excited about that for some reason. So, I’m going to eat well, I’m going to be warm, and I’m going to go anywhere I want and use the AI tech.

So, I think I’ve got grand ambitions.

Matt Patrick: Yeah, that all sounds great. I mean, robotic chef. I feel like I need to follow a podcast on that specifically. I have two main takeaways for that. One, you clearly didn’t return the Oculus Rift that you got for your son. You kept that for yourself for retirement. So that makes sense to me.

And then it feels like your move is you’re going to start talking to your kids, ”Hey, Arizona and Florida, those are great places, because your mom will come with us if you’re down there.” So you got to start working on them.

Mike Allen: You got my strategy down. That’s exactly what I’m going to do.

Matt Patrick: Beautiful. Well, Mike, we really appreciate the time. Thank you so much for hopping on and chatting with us. And I think this is a lot of great takeaways for us and for everybody that listens. Thanks so much for your time.

Mike Allen: No, thank you. I really enjoyed being on the podcast.

Thank you both.

All right, Dawn, a lot to digest there. What are your initial reactions?

Dawn McPherson: My initial reaction is this topic is way above my head. I find it fascinating and there were so many little offshoots I wanted to go down. Like I said, I was going to circle back on the real-life application and I never got to do that, but it’s okay because there was so much great content. I think this is our first time on a podcast focused on the topic of AI. I can imagine that we’ll revisit this topic periodically because it’s just going to continue to evolve probably at a pretty rapid pace.

My favorite thought about AI is the idea that it could fix what I’m saying while I’m on presentations.

Matt Patrick: “What you’re saying didn’t make any sense. Want to try that again?”

Dawn McPherson: What about you? I have other thoughts, but I would love to hear your initial thoughts.

Matt Patrick: I’m with you on the scope of everything that’s in there. I’m with you on processing the notes afterwards. I feel like we could do specific follow-ups on all the different avenues this could go down.

Part of what we’re trying to highlight from a plain sponsor perspective is all the different avenues where you could see impact here. It could be direct to your employees. It could be at your level in terms of interacting with your provider or your asset manager. Could be behind the scenes, like, Hey, this provider’s website got a facelift and it works well. What happened there?

With the developers we had, we could do two times the work and it’s flowing well. I think trying to keep that in mind as we continue to monitor it. There’s all these different areas where it could be. And I think sometimes we jump ahead to Oh, we’re all going to be talking to AI bots and that’s how we’re going to interact with everything.

Maybe that’s not step one. Step one is going to be some of these immediate items that maybe you dislike doing. Someone’s going to build a tool that makes that easier. And it’s going to feel more streamlined and less like you’re in a sci-fi show.

Dawn McPherson: And that’s exactly what we start to think. I think that’s where our minds go. I shouldn’t assume, but I think that’s where a lot of our minds go is like you just said, some type of sci fi show. But think back however many years, would you have ever thought we’d be typing notes to each other on a phone instead of having full-on conversations?

It’s just the reality of where we’re going. And I think if you can take it in those little digestible pieces and think about, Hey, this is just a tool. Well, it’s not just a tool, but if you think of it this way, maybe, I don’t know. Yeah. It’s a tool that can help me in these various pockets of my personal life, but also in our business.

There’s going to be so much application. And I think I mentioned this in our prep. Matt, but talking about we just don’t know what we don’t know. And sometimes it takes someone else’s creativity to think of uses for new systems and technology. And so I think in six months, even, we’re going to see so many other things being done with AI.

Obviously, there will be some bad apples in there using it with bad intentions or what have you. Those are just some of my initial thoughts.

Matt Patrick: Yeah. I think that’s right. And I think it’s also a good thing for us to bookmark and say maybe this should be a topic we follow up on a few months from now. It’s probably something we just keep an eye on moving forward on this podcast feed, just to see what’s coming down the road, what’s being implemented, what’s the timeline for something. Because I do think it’s going to get a lot of buzz. And so filtering through some of what’s noise and what’s actually impactful today will be something worth watching.

Dawn McPherson: Yeah, I agree. Good segment. I did enjoy your Buy Sell segment creation. That was fun. I might’ve needed to be a little clearer on buy means agree.

Matt Patrick: Well, I think like we’re saying with AI, like there’s going to be some bumps along the way. We tried something new. That’s just how it goes.

Dawn McPherson: That’s where I needed the AI to help me tee up that segment, right?

Matt Patrick: Yeah. No one knows what you’re trying to do here.

Dawn McPherson: All right. Well, we’ll call that a wrap on today’s revamping retirement. Thanks so much to our listeners for joining us. Be sure to like and follow us on whatever platform you choose to listen to your podcasts on. And with that, we’re going to turn it over to our Minute with Mike.

Mike Webb: This month we will discuss how ERISA retirement plan sponsors can improve a process that is a source of dread for many, the filing of 5500 annual reports. The first area of improvement, at least for those plans with 100 or more participants, is taking control of the audit process to which such plans are subject.

The number one reason that 5500 is delayed is the audit. Plan sponsors should debrief their prior audit process now. Explain to the auditor the goal for the 200,000 023 filing, or filings, is to prepare as though the filing is actually due on the non-extended deadline date of July 31. It is not impossible.

The second area of improvement is for plan sponsors to get a handle on the number of plans they maintain. Many employers sponsor just one ERISA retirement plan, while others have two, three, or even five or more plans. Each additional plan can greatly complicate the 5500 process, never mind the fact that it also increases the overall reporting and disclosure burden under ERISA. Worse yet, employers who sponsor multiple plans often do not have a legitimate business reason for doing so.

Thus, plan sponsors should work to consolidate plans, or if that is not feasible, to explore the new group of plans option to determine if a single 5500 filing should be completed for multiple plans.

The final area of improvement is for plan sponsors to keep the lines of communication open with all service providers involved in the 5500 process.

If there are significant changes to a plan during the year, for example, a recordkeeper change, trustee change, or a change in anything else that is mentioned in the plan’s audited financials, October is not the time to be informing the plan auditor or 5500 preparer. Service providers should be kept in the loop as changes occur.

If the plan sponsor can take these simple steps, this year’s 5500 process might actually be a pleasant one.

The discussions and opinions expressed in this podcast are those of the speaker and are subject to change without notice. This podcast is intended to be informational only. Nothing in this podcast constitutes a solicitation, investment advice, or recommendation to invest in any securities. CAPTRUST Financial Advisors is an investment advisor registered under the Investment Advisors Act of 1940.

CAPTRUST does not render legal advice.

Thank you for listening to Revamping Retirement.