How Marriott Tackles AI, Complexity, and Fake Candidates

What happens when you mash up AI, a million hotel workers, and the fourth-largest restaurant chain in the world (yes, that’s what Marriott technically is)? You get Tyler Weeks, Managing VP of Tech & Analytics at Marriott, trying to wrangle HR chaos that makes Outback Steakhouse look like NASA.

We dig into:

• Why Ritz-Carlton still won’t take Joel’s reservation (seriously, Tyler, explain).

  
  

• How Marriott’s HR looks less like a corporation and more like the Little Rascals in a trench coat trying to get a bank loan.

  
  

• Why AI doesn’t make rockstar recruiters better, but it does turn your B-team into something resembling the A-squad.

  
  

• The ROI fantasy math CFOs pretend to believe (“Sure, this software will save us a billion dollars… right after I sell you this bridge”).

  
  

• And yes—fake résumés, deepfakes, lazy applies, and how the ultimate fix is… wait for it… having candidates show up in person. Revolutionary.

If you’ve ever wondered how to herd 9,000 hotels, 40 languages, and a tsunami of fake applicants into one semi-functional talent machine—this is the AI Frontline you don’t want to miss.

PODCAST TRANSCRIPTION

Joel Cheesman: All right, let's do this. We are The Chad and Cheese Podcast. I'm your co-host, Joel Cheesman. Joined as always, Chad Sowash is in the house, and this is the Sessions AI Frontline Series as we welcome Tyler Weeks, Managing VP of Technology, Analytics, and Research. Tyler, welcome again to HR's Most Dangerous Podcast.

Chad Sowash: He's back.

Tyler Weeks: I'm so excited.

Joel Cheesman: So let's get down to business. Why won't the Ritz-Carlton take my reservations?

Tyler Weeks: That's a good question.

Joel Cheesman: All right, we'll circle back to that. For our viewers out there that don't know Tyler Weeks, give us the elevator pitch.

Tyler Weeks: The elevator pitch. Well, what I can tell you is, so I work for Marriott International, and I've been with them about three years. And it is a scale of a problem like I've never tackled before.

Chad Sowash: Talk about that.

Tyler Weeks: It's a lot of fun. So we've got over 400,000 employees worldwide. We're in 140 countries. The languages that we have to translate for, even just to issue a survey in the US, I have to translate about 40 languages, just given our workforce.

Chad Sowash: Hello.

Tyler Weeks: If you took, and I can't verify this, but I've heard from multiple sources internally, that if you took just our restaurant business, we'd be the fourth largest restaurant chain in the world, which I don't know why being the biggest hotel doesn't sound as impressive as being the fourth biggest restaurant chain, but in my head, it does.

Joel Cheesman: Because your experience at Outback is much better than your experience at the average hotel.

Tyler Weeks: Than at Ritz-Carlton.

Joel Cheesman: Not Ritz-Carlton, because I haven't been there, Tyler. We'll get back to that. Put a pin in it.

Chad Sowash: He loves an Alice Springs Chicken.

Tyler Weeks: So we're huge, and that's only half of our business. So the other half, we also franchise. And if you take the franchise side of our business, we're closer to a million people wearing our badge around the world. And so if you're in HR and you're worried about their experience, about their training, about onboarding, hiring, recruiting, the scale of that problem is a lot of fun.

Joel Cheesman: How big is the TA department in and of itself?

Tyler Weeks: That's a... It's huge. Well, what I can tell you is that we've got about 8,000 HR professionals worldwide.

Chad Sowash: There's that.

Tyler Weeks: And that doesn't include sort of outsourced BPO or sort of business process support that comes in with TA and with benefits and some other.

Joel Cheesman: It's big.

Tyler Weeks: It's huge. It's a big operation.

Chad Sowash: You're all over the world. And they're just all the same business. So I'm sure it's just one standard process. Everybody uses the same stack. It's easy, right?

Tyler Weeks: I like to joke that we're not really a business. We're like the little rascals stacked up under the overcoat, like doing an impression of an adult trying to get a loan at the bank. Because we're, there's like...

Joel Cheesman: I get froggy vibes from Tyler. I don't know about you.

Tyler Weeks: We've got, you know, almost 9,000 hotels. Each one is kind of a medium, small to medium-sized business on their own. And some of those hotels have 2,000 employees. So they can be a very large operation. I mean, if you think about, if you're familiar with like the Gaylord Resorts in the US.

Chad Sowash: Yeah, Joel can't go there either.

Tyler Weeks: Huge. Huge.

Joel Cheesman: Some with the Hee Haw girls back in the day.

Tyler Weeks: Huge operation, right? And so, you know, the HR team there alone is a large team in its own right. And so they've got unique challenges. They've got owners and investors that are part of managing that hotel that have some say in how we operate. So we try to drive as much uniformity as we can for scale. But we do have to respond kind of at a very granular level. If you, just at the risk of ranting here, if you think about from like a TA perspective, our talent that we're competing for isn't just other hospitality companies.

Joel Cheesman: It's yourself.

Tyler Weeks: It's ourselves. We might have two hotels across the street or like hospitals, right? Hospitals are a lot like a hotel, just worse beds and more beeps. But, you know, like you've got a lot of the same.

Joel Cheesman: Worse food too. Yeah.

Tyler Weeks: Worse food. You got a lot of the same roles or even an Amazon fulfillment center going in down the street can pull a lot of our hourly talent away. And that's based just on commute, not on like, I can't ship housekeepers across the country.

Joel Cheesman: But Tyler, I read headlines all the time. This is all getting AI'd. There are robots going to fold, clean my towels. Is that not... Am I not reading the right headlines? Like what is automation and AI meant to this whole quagmire of an organization that Marriott is?

Tyler Weeks: The next Roomba is going to make your bed. Look, that is actually one of the most sort of exciting sort of fun parts of thinking about how we operate. And from a research analytics standpoint is kind of fun because, you know, I've worked at like software or hardware companies before, tech company. And when you're a large monolith of a company, you don't have a lot of opportunity to experiment, right? Because you're sort of looking at one big entity and trying to figure out how it operates. Within a hotel company, we're kind of like, it's more like sports statistics because I've got all of these repeating units that have a lot of the same positions. And what's different is their context, like where they are, the brand, the size, the political climate, all those things. And so we can look at all kinds of stuff about how those hotels are operating. And some of the things that the folks doing research on my team are looking at are some of that, like that interplay between how you even bundle roles together. Like, does it make sense to still have a bellhop be separate from concierge, be separate from eight other roles that are kind of doing similar things?

Chad Sowash: Yeah.

Tyler Weeks: And then how you fold AI into that and how you really help your front desk talk to people like they know you. Welcome back, blah, blah, blah. All that sort of personalized experiences is some exciting work that's happening kind of top to bottom.

Chad Sowash: Well, being able to feed those large language models, you've got to have structured data. Is there a way that you can actually do that in aggregate or do you do that country by country? How does that, actually, how do you try to roll something out that big that's so compartmentalized?

Tyler Weeks: Yeah, actually, that's where I feel like one of the... There's been so much focus on with the new AI models around chatbots, I think mostly because it was called ChatGPT. So it just sort of set the context and we just talk about chatbots. That and prompts. That's kind of dominated the conversation around those two things. I think the less well sort of debated or in solution problem is knowledge management. It's always been a problem in HR, like documenting in TA, like documenting all your processes, local variation, making sure that it's actually real, like represents how people work. This makes it twice as, 10 times more critical because if somebody is going and asking, I need it to give the right answer and I need to give the right answer in Brazil, not just kind of a generic answer that's sort of roughly true everywhere.

Joel Cheesman: The nuance of the location.

Tyler Weeks: Exactly.

Chad Sowash: In data and not anecdotes.

Tyler Weeks: Data and not anecdotes. So it's connecting to data sources. So we are actually, I just formed this year a new team that we call process transformation and it sits side by side with my technology team because I see the work that they do is just as critical to the technology effort, engineering both the processes and the document, how we store that documentation. So we've started this effort to not only create a knowledge library that we could take any model, AI model and put it on top of, that it could answer quite, because I want to stay on top. I want to always have the best one and so I need it to be sort of like plug and play. Like I've got the library.

Chad Sowash: Well, testing all the different models and then being able to switch in and out.

Tyler Weeks: Right. And so if I've got all of my knowledge architected in a way that I could swap in any model anytime and have it be ready to go day one. So we've kicked off an effort to do that and what we're doing is we're building in how we're structuring it, places for local teams to add their documentation. So how do you modify the TA, like how is TA different in Austria than in the US or in California? Because even within the US, state to state, there's so much nuance. This is going to take us years to do, but I think it might be the bigger technology problem is knowledge management.

Joel Cheesman: Tyler, this sounds really expensive and complicated, but what I'm hearing from you is this is a cost savings. Talk about that.

Tyler Weeks: Yeah, I mean, it's certainly an investment. I've been in this space now for about a decade, which is crazy to me. I still feel like a rookie in the HR world and I'm going to use the new guy card for 20 years.

Joel Cheesman: That's over.

Tyler Weeks: The promises of 5, 10 years ago of AI and TA, or HR more generally have always been more of like a soft dollar, soft savings. It's like we're going to give you time back for your recruiters to have those meaningful conversations. We're going to give you time. Time doesn't do me any good. And the reason is, there's a paradox in that. You've probably heard of it. Parkinson's law says that the time to complete a task will expand to fill the time allotted. Right. You've heard this in kind of a joking terms, but it's real. You can automate somebody's work or part of it, they will feel just as busy. So here's the irony is that if I spend a bunch of money to automate half your job and I'm successful in doing that, but I haven't actually changed what you do. I've made you less productive because now you're doing less per hour.

Chad Sowash: Yeah.

Tyler Weeks: And I'm spending for your salary and the software and you're doing less work.

Chad Sowash: So you have to formulate on both sides of it.

Tyler Weeks: Both sides. Yes. So what I've started to say is that I feel like the place to innovate and the place that I really see my contribution to the company is I want to hand time back to my employees and I want to hand dollars back to my company. And I think that HR sits in a place to really drive that. And I think some of the places that's going to happen in the future, and this is just us, I think in general is higher functioning sort of shared service models. Because the research that I've seen around these new AI models is that they don't tend to improve performance of your highest performers.

Chad Sowash: Right.

Tyler Weeks: They know their stuff. You take your best recruiter AI isn't making them significantly better. They're good. They know how to connect with people. They close offers. But what it does do, it gives a disproportionate bump to your lowest performing recruiters or your new recruiters, people earlier in their careers or less familiar with the field because it equips them with what to say. That's magic.

Joel Cheesman: How close can it make them to the top level recruiter?  Like, what's the percentage difference? 

Tyler Weeks: I think they're. So the research I saw was more around call center. Some experiments with a call center. And I think I'm going to misquote this, but I think they saw like a 30% productivity bump on your low end and almost zero on the top end, which is kind of backwards with how it gets talked about, right, and experimented with is its got to give your top performers the AI and be like, okay, find some efficiencies.

Chad Sowash: So we should just be hiring B players and equipping them with AI to give them truth to that?

Tyler Weeks: Absolutely. And I don't mean B players as in they're not ambitious or they're not good. But I need to get a critical mass of people that are supporting TA and other HR functions. I can get more creative in where I look and what education levels I need. And that's where I think I've got a big opportunity around handing dollars back to the company and handing time back to our employees.

Chad Sowash: So on the ROI side of the house, there are many different points, it seems like, of light that you can start to pull instead of soft savings. You can start to demonstrate perspective, hard impact on bottom line. Is that something that, I mean, you're trying to formulate to be able to not just... Time means something, but generally it also means something with regard to getting somebody in a seat faster because that seat that has nobody in it, there's no productivity, right? So what about that aspect of it as well?

Tyler Weeks: That one has always been... If you're making a case to a CFO or your finance controller about investing in software, that one's always a sticky one because I can multiply hours and take assumptions, and no matter what assumptions you pick, I can save the company a billion dollars with onboarding faster. It just multiplies in a way that makes it look like it's a slam dunk. I should be able to buy any software I want based on that.

Chad Sowash: Doesn't work that way.

Tyler Weeks: It doesn't work that way. It's hard because cost avoidance isn't nearly as powerful as cost reduction.

Chad Sowash: Yeah.

Tyler Weeks: And the way onboarding, like that time to productivity, often gets sort of accounted for is more like a cost avoidance. Like I brought somebody on and they were less productive and I avoided that lack of productivity. So it's a tough thing. You can do it and people do do that successfully often, but you really do have to look at reducing software costs overall. I do see as platforms have gotten more mature over the last 10 years, really the HR tech space has consolidated in a type of way, not really around specific vendors, but around bundles of capability.

Joel Cheesman: Features are all looking like each other.

Tyler Weeks: Yeah. Can I name names?

Joel Cheesman: Sure. Of course.

Tyler Weeks: If you told me six years ago, seven years ago that Paradox, HireVue and Phenom would basically all have the same features, I would have...

Joel Cheesman: Laughed you out of the room.

Tyler Weeks: I couldn't have conceived of a world where they would like...

Chad Sowash: But it's convergence though, right?

Tyler Weeks: Yeah. They're kind of converging on a similar thing. I think what, and with these agents that are coming out now, I think what you're going to start seeing is more companies sort of finding a happy medium between best in breed and enterprise where you've got a good foundation that's enterprise-centric and then you've got overlays or plugins that are best in breed where they're going to make the most strategic difference.

Joel Cheesman: I want to go back to your breadth of the organization and the amount of stuff from the job seeker side that you see, target, eliminate, whatever. Lazy apply, deep fakes. We hear so many stories about on the job seeker side how it's a tsunami of stuff.

Chad Sowash: It's scary stuff.

Tyler Weeks: It's real. Yeah, it's real.

Joel Cheesman: So you see it all all across the globe. What are you seeing in terms of what job seekers are doing and how are you combating some of the malpractice?

Chad Sowash: The Jason Voorhees of AI.

Joel Cheesman: Yes.

Tyler Weeks: You know what's... I wish I had more on this around like a real trend or real research but I can speak anecdotally like internally because yeah, just like anybody else, we've started to get this uptick in fake applicants. Or sort of, it's a real applicant but there's sort of this out there, like a front for this outsourced thing where there's like you know you've seen the headlines around all that stuff. You know what's kind of like the lo-fi thing? It's just in-person interviews.

Joel Cheesman: Sit in by a Kinko's and a FedEx package.

Chad Sowash: How do you break that? Just have them come in.

Tyler Weeks: Just talk to them.

Joel Cheesman: Have them come in.

Tyler Weeks: Yeah. You just have them come in. I think that's...

Joel Cheesman: So you guys are doing more of that?

Tyler Weeks: We are doing more of that. Yeah. We are doing more of that. We do have more people than, you know, we kind of think hit a high point during COVID like everybody of doing everything remote, everything online and I think we're seeing a bit of a correction of that even in our own practices around like, it's hard to deep fake it.

Chad Sowash: Yeah.

Tyler Weeks: Although I am a hard...

Joel Cheesman: So aside from the Marriott and the Ritz in North Korea, all the other operations are suffering from that. We'll leave it at that. Tyler, thanks for hanging out with us.

Tyler Weeks: Yeah.

Joel Cheesman: This has been The Chad & Cheese Podcast. For Chad Sowash, I'm Joel Cheesman. This has been the Sessions AI Frontline series. We out.

Chad Sowash: We out.