AI-Powered Travel Brokers: MIT and IBM Revolutionize Trip Planning with Personalized AI

Navigating the complexities of travel planning, from transportation and accommodations to meals and lodging, can be daunting. While large language models (LLMs) hold promise for simplifying this process through natural language interaction and information gathering, they often struggle with the intricate logistical and mathematical reasoning required. A recent study revealed that state-of-the-art LLMs provide viable solutions for complex trip planning only 4% of the time, even with additional tools and APIs.

To address this challenge, a research team from MIT and the MIT-IBM Watson AI Lab has developed a novel framework that significantly enhances the success rate of LLM solutions for complex problems. Chuchu Fan, associate professor in the MIT Department of Aeronautics and Astronautics (AeroAstro) and the Laboratory for Information and Decision Systems (LIDS), explains that they reframed the issue as a combinatorial optimization problem, where multiple constraints must be satisfied in a certifiable way.

Recognizing the transferable nature of their work, the team created a user-friendly AI travel broker that develops realistic, logical, and complete travel plans. This framework combines common LLMs with algorithms and a complete satisfiability solver. Solvers are mathematical tools that rigorously check if criteria can be met and how, but they require complex computer programming for use. By integrating solvers with LLMs, the researchers enable users to plan trips efficiently, without programming knowledge or extensive research. The technique can identify and articulate issues if a user’s constraints cannot be met, proposing alternative measures for a valid plan.

“Different complexities of travel planning are something everyone will have to deal with at some point. There are different needs, requirements, constraints, and real-world information that you can collect,” says Fan. “Our idea is not to ask LLMs to propose a travel plan. Instead, an LLM here is acting as a translator to translate this natural language description of the problem into a problem that a solver can handle [and then provide that to the user].”

The research, co-authored by Yang Zhang of MIT-IBM Watson AI Lab, AeroAstro graduate student Yilun Hao, and graduate student Yongchao Chen of MIT LIDS and Harvard University, was recently presented at the Conference of the Nations of the Americas Chapter of the Association for Computational Linguistics. Their method uses satisfiability modulo theories (SMT), which determines whether a formula can be satisfied and reasons over different algorithms to understand whether the planning problem is possible to solve, considering all limitations and constraints.

The “travel agent” operates in four repeatable steps, using GPT-4, Claude-3, or Mistral-Large as the method’s LLM. First, the LLM parses a user’s requested travel plan prompt, noting preferences for budget, hotels, transportation, destinations, attractions, restaurants, and trip duration. These steps are then converted into executable Python code (with a natural language annotation for each of the constraints), which calls APIs like CitySearch and FlightSearch to collect data, and the SMT solver to begin executing the steps laid out in the constraint satisfaction problem. If a sound and complete solution can be found, the solver outputs the result to the LLM, which then provides a coherent itinerary to the user.

If one or more constraints cannot be met, the framework begins looking for an alternative. The solver outputs code identifying the conflicting constraints (with its corresponding annotation) that the LLM then provides to the user with a potential remedy. The user can then decide how to proceed, until a solution (or the maximum number of iterations) is reached.

Testing the method against other baselines, the new technique generally achieved over a 90 percent pass rate, compared to 10 percent or lower for the baselines. The team also explored the addition of a JSON representation within the query step, which further made it easier for the method to provide solutions with 84.4-98.9 percent pass rates.

The MIT-IBM team also applied their framework to other domains with tasks like block picking, task allocation, the traveling salesman problem, and warehouse optimization. Hao summarizes, “I think this is a very strong and innovative framework that can save a lot of time for humans, and also, it’s a very novel combination of the LLM and the solver.”