TAKO: an automated design thinking process facilitation proof-of-concept chatbot

The primary objectives of this project were to design and create a chatbot that supports students in navigating and selecting suitable design thinking methods, train the chatbot to improve its eff...

The primary objectives of this project were to design and create a chatbot that supports students in navigating and selecting suitable design thinking methods, train the chatbot to improve its effectiveness, and provide students with fewer opportunities a chance to enhance their design thinking knowledge. Additionally, we aimed to disseminate the developed chatbot and its results. Due to the advent and integration of LLM technologies, we extended our original objective to include the development of a general-purpose design thinking chatbot. This chatbot now supports students and teachers by addressing both general and specific questions about the design thinking methodology and various design methods.

Throughout the project, we implemented a series of structured activities. We began with the chatbot's design, conducting extensive research and user interviews to inform its development. We then moved on to creating and validating the chatbot concept through initial testing and simulations. The development phase included integrating relevant APIs. In the second iteration of the chatbot, we completely switched to an LLM-based chatbot. Training the chatbot involved creating a verifiable knowledge container on Design Thinking resources and information and refining the chatbot's responses and capabilities. Lastly, we conducted real-life testing with students and teachers, gathering feedback to improve the chatbot further. Dissemination efforts included creating a result and recommendation report, presenting at educational fairs, launching a dedicated website with the chatbot instance, and hosting online events to showcase our work.

The primary outcome of our project is a functional and freely available chatbot prototype that supports design thinking practices. We developed both an online working version of the chatbot for teachers and a publicly accessible code database for further improvement by other educators and coding enthusiasts. The prototype exceeded our initial objectives, with over 270 students interacting with the chatbot and more than 1500 requests processed. We iteratively improved the chatbot through three complete versions. Additionally, we engaged with over 15 schools, primarily from Belgium and the Netherlands. Based on the chatbot's performance evaluation, we also created three future scenarios for its improvement and effective use in educational practices.