Context: 

Humans are interacting ever more with machines. The recent past shows that so-called “AI/HI ensembles” – collaborations between Artificial and Human Intelligence – can be powerful indeed: the best chess players are humans supported by machines, and large language models significantly improve productivity. 

To maximize their potential, however, mutual understanding between human & machine is needed. 

Project objectives:  

The goal was to investigate the use of chatbots to allow regular business users to interact with their data to perform advanced analytics. The project implemented a generic methodology to develop domain-specific chatbots based upon a detailed description of a dataset. 

Description of the work done, and the main results obtained:  

We extended AI models to make them interactive and explainable. This enabled business analysts to intervene during their construction and understand their inner workings. We explored new frontiers in the domain of HI/AI where intelligent systems are co-created by man and machine. 

We implemented an experimental chatbot capable of interactively extracting valuable insights and building advanced personalized machine learning models through conversation.

Research partners: ErasmusHogeschool