As artificial intelligence (AI) continues to dominate headlines and garner more hype, travel companies have started hiring for AI-specific positions, even altering their corporate structure to add executives. While some brands have created positions more recently, others have had AI in mind for some time.
Michael Wu is the chief AI strategist for PROS—and has been since 2018.
Wu works cross functionally with the research team to prototype the company’s advanced algorithm. Additionally, he collaborates with the product team to incorporate more AI features into PROS’ product. Outside of PROS, Wu is an academic, working as an instructor at the University of California, Berkeley and in research at École des Ponts Business School in Paris.
In an interview with PhocusWire, he discussed his role and his thoughts about AI, airlines and the future of the industry.
Read the full discussion below:
* Editor’s note: Answers have been edited and condensed for clarity and length.
Tell me a bit more about what your role entails. What's your day to day like with PROS?
It can be very different. Sometimes you can be just working with the research team, discussing what data transformation is needed to drive prediction accuracy—that's more of a research scientist kind of work and analysis. And then there's also product discussions regarding what type of agent can we build … I work with the product team as well to refine the product.
Subscribe to our newsletter below
And then sometimes it could be speaking at conferences, giving workshops for customers—and then the internal initiatives are working cross-departmentally to see what AI can we use internally and if there are any overlapping efforts across different departments so can we consolidate some of these efforts.
Why is there a need for a chief AI strategist right now?
I think every company will probably have some kind of chief AI strategist. Depending on the company, some of them may have one or two different roles. Some could be more research-oriented, some may be more internal-focused and some will be more external-focused. The reason is because AI is actually changing the way we work.
How has your role changed since you started in 2018?
The big change actually comes from this wave of generative AI. ChatGPT basically popularized generative AI. In a lot of respects, it made the job easier, because we don't need to convince people that AI is something important. Everybody knows. Everybody wants to learn about it.
I think people still need to be educated and understand how AI can be used most effectively.
How would you advise the travel industry to be thinking about AI right now?
Don’t be afraid of experimenting with AI. I think a lot of people are so risk-averse that they are not willing to do any experiments or pilots or anything. That’s just going to put you so much further behind. People who start earlier, they have a first-mover advantage and they are going to go much further.
And get your data ready. I think a lot of big organizations, like the airlines, their data is actually spread all over the place. The thing that we know about AI is that it is fueled by data. If you don't have data, then the AI engine just won't run, the algorithm won't do you any good. Get the data governance ready so that AI can use that data to help you make better decisions.
How big of an issue is data fragmentation? How problematic is it for airlines?
Within the airline industry, there are platforms that are actively limiting access to their data. That’s slowing and hindering the industry from moving forward. I see that as a terrible thing—airlines have a lot of obviously inherent advantages in the sense that it is a very data-rich industry. But if you don’t let people use it, don’t let AI use it effectively, then you’re not actually getting the benefit from having the data. So I think that’s a big challenge.
ChatGPT basically popularized generative AI. In a lot of respects, it made the job easier, because we don't need to convince people that AI is something important.
Michael Wu, PROS
Hopefully, as we develop these AI agents that can actually help you get the data, that process will become easier, but that still needs to be allowed by respective platforms.
What's the best way to address data fragmentation that would most benefit airlines?
There are many ways I can tell you how some of our customers are solving this. One of our customers is actually pulling in data from disparate sources into a central data lake. And then from there, your AI algorithm can basically get visibility into all kinds of data based on what kind of credential it actually is provided with. So this credential provided access to these five different data sources that all got pulled into this data lake. It normally would have to go to find different sources, and it would be crazy and the integration can be very brittle, and things will not work and these algorithms will not work.
One of our customers, they are actually simply pulling data nightly. This is pulling all this data from all the different disparate systems into a central data lake. From there, the AI algorithm can actually take advantage of having visibility to a wide variety across departmental data.
That's within one airline, correct? Is there data sharing that goes on between legacy airlines?
Well, it depends on what kind of data, right? I think cosharing or data sharing is definitely possible. But airlines are actually generally very sensitive about sharing their data. For example, like the forecasting of bookings, that data is extremely confidential. They actually are very unwilling to share those data with others; they see that as a competitive advantage.
What are the other biggest roadblocks for airlines as they consider AI implementation?
Besides the data, I think it is just the inherent complexity around the multiple systems. I think a lot of those actually center around data, but some part of it actually has to do with governance.
And if there's one other thing, I would say the risk aversiveness. Airlines are just extremely risk averse. Airlines could benefit a lot if they could just embrace this experiment. Because this is new technology. The faster you could actually experiment with it, the faster you can iterate, [which] means you could iron out these potential risks faster.
To what degree are legacy systems or legacy technology becoming a roadblock to success with AI?
The fragmentation of technology is a big problem, and the concept under that is the data is fragmented.
The alternative is that you create some kind of agent that basically uses this legacy system as a supporting tool. We can actually teach them how to call these traditional systems as tools so that their response will be grounded, and they'll be grounded by these traditional tools.
How do you see AI changing the travel industry? How much of this is hype and how much is reality?
It's definitely a reality. I think today, people are definitely using their language model, like for suggestions or trips itineraries and stuff like that, travel tips and all that.
Today, if you just ask ChatGPT [for an itinerary], it will give you an answer. It may not be perfect, but you probably won't notice, because there's not a right answer. But if you're asking, ‘What's the price or ticket for certain travel?’ There is a right answer. There is an optimal answer. And that's what a lot of these traditional systems, reservation management systems, provide. Those are the ones the AI needs to learn to leverage—those existing systems, those tools—to ground their response before answering the user.
How do you see agentic AI changing the travel industry?
Agentic AI is probably more important on the business operation side. It can definitely help airlines operate much more efficiently.The agents can automate some of the work that is normally done by humans.
I think it’s going to make internal operations of an airline much more efficient.
What do you think is next on the AI front, particularly for travel and especially for airlines?
They call 2025 the year of the agent because everybody is doing something with an agent. I think in some industries where agents are adopting more, probably they will move beyond. [In] the airline industry, this will probably still be a big trend in the next year or two.
What's your wildest AI-related prediction for the travel industry over the next five years?
I think it could potentially completely change the flying experience. I think today, if you ask a lot of people how their trip is going … people often say that the flying part is the least enjoyable. But I think that that could change. I think if you know you have the proper infrastructure—that's the type of transformation AI can bring.
Tell Phocuswright how your company is using AI
Phocuswright is conducting a survey to understand how travel companies are using AI, including its impact on operations, marketing and service. Complete the brief survey for a chance to win a ticket to The Phocuswright Conference in November.