Artificial Intelligence for Handling Cancelled Flights: VŠE Joins a Project with Kiwi.com Labs
A cancelled flight or a change in travel plans means having to quickly assess the available alternatives, the carriers’ conditions, and the options for a refund or compensation. These are exactly the situations addressed by the project “Neurosymbolic Platform with a Context Engine for Autonomous Process Management.” The Faculty of Economics of the Prague University of Economics and Business (VŠE) is taking part in the project together with the VŠE Faculty of Management in Jindřichův Hradec. The lead member of the consortium is Kiwi.com Labs, and the other partners are FT Park and ICON.
The aim of the project is to develop a platform that combines the ability of language models to understand customer requests with precisely defined decision-making rules. This so-called neurosymbolic approach is intended to let the system draw on up-to-date data about the booking, flight availability, and the applicable conditions when handling a specific case. It will also be important to be able to trace back and explain why the system chose a particular course of action. The platform is expected to gradually handle the entire process, from assessing the situation, through proposing a solution and communicating with the customer, to carrying out the necessary steps, with the option of handing the case over to a human.
The joint involvement of both VŠE faculties brings an economic and managerial perspective on the use of artificial intelligence to the project. Together with its partners, VŠE will help analyse decision-making processes, compare different approaches to automation, and assess how clear and traceable the system’s decisions are.
The work will also include experimental testing of how users react to various scenarios for resolving travel disruptions. This will help show how well users understand the proposed solutions and how they respond to them. A significant part of the work will be measuring actual costs and benefits. The researchers will look at how the time needed to resolve customer requests changes, as well as the volume of manual work, the error rate, and the frequency of cases that have to be taken over by an operator. The evaluation will also include the cost of running the AI itself and of correcting wrong decisions. Through its involvement in the project, the Faculty of Economics will gain an opportunity to apply economic research to concrete problems in digital services.
Interview with Ing. Josef Klement, Ph.D., economics specialist on the research team:
What was it about the topic that interested you so much that you decided to lead the project on behalf of VŠE?
What appealed to me most was the chance to collaborate with such a successful Czech technology company. That doesn’t happen every day. I also enjoy taking on challenges, and combining AI development with economic research and experimental testing of user reactions is certainly one. The project will also let us study situations that many of us know from our own experience. When your flight is cancelled, you need to find out quickly what your options are and what each solution means for you. This is exactly where it will be interesting to observe how people understand the AI’s suggestions and how they make decisions based on them. Together with the Faculty of Management in Jindřichův Hradec and the other partners, we can examine under what conditions automation will make resolving travel disruptions easier while also making economic sense.
Which part of the research do you personally consider the most challenging, and which part are you most looking forward to?
I think the most challenging part will be designing the experimental scenarios. They need to faithfully capture the situations travellers deal with, and they also need to let us distinguish what influences their reactions. When travel plans change, a whole range of factors can play a role, from time pressure and uncertainty to, for example, the way the system presents the available options. The testing of user reactions is what I’m looking forward to most, because it will reveal how understandable and suitable the proposed solutions are for users. Perhaps thanks to this, interacting with the emerging platform will also be more pleasant and calmer in the future.
Do you see other directions in which this research could develop in the future?
Definitely. The project already anticipates extending the platform to situations where, for example, a large number of flights are disrupted at the same time. From an economic point of view, it will be interesting to look for the right degree of system autonomy, taking into account costs, risks, and user reactions. In the future, the insights we gain could also be applied to other services that combine complex rules, rapidly changing information, and the need to respond to a customer’s individual situation.
What will the project bring to the Faculty of Economics, its researchers, and possibly its students?
Kiwi, the company to which Kiwi.com Labs s.r.o. belongs, is one of the largest and most successful Czech technology companies, and this collaboration opens up a whole range of opportunities for us. We can apply economic research and experimental methods to the development of advanced digital services. We can publish academic papers based on our analyses and economic experiments. We can develop the collaboration further, build contacts, and prepare more joint projects. We can bring our findings from the project into our teaching and inspire our students. There are many opportunities, and I’m glad that our faculty and all of us can benefit from them.