Conference: Environments for Collaborative Knowledge Construction, Large Language Models (ChatGPT, Flaubert), and Small Discrete Models (JeuxDeMots, Hélix, Idéfix)

As part of the Master's in Computer Science graduation ceremony, the FdS Department of Computer Science is hosting a lecture open to the general public.

Friday, June 2 – 2:30 p.m.
, Dumontet Lecture Hall

Lecture led by Mathieu Lafourcade, Professor at Lirmm

It is interesting to develop environments for the co-construction of knowledge between human experts and one or more artificial intelligence (AI) systems. Human experts discover knowledge with the help of AI, and the AI systems learn (primarily from common sense). A conversational interface, such as a chatbot, seems appropriate because it is intuitive and versatile. We are exploring the use of ChatGPT (based on GPT 3.5) and another language model, Flaubert, and are attempting to assess their performance in this context using the JeuxDeMots lexico-semantic corpus.

We also present a discrete neural model that serves as an overlay for the JeuxDeMots lexical network, which can be viewed as an illustration of a different approach. This model employs forward-forward learning and appears to be a biologically plausible and effective alternative to backpropagation. The architecture of this model is not fixed a priori and evolves over the course of training through the dynamic addition of neurons.

To illustrate our point, we present the OACS project, which aims to automatically determine whether certain clauses in a consumer contract are unfair.