Conversation on AI: past, present, and future

On the seventieth anniversary of the historic Dartmouth conference, where the term “Artificial Intelligence” was introduced, the conversation between Stefano A. Cerri and Enrico Motta will trace the evolution of AI, starting from the “General Problem Solvers” of the sixties, moving through knowledge representation, ontologies, and intelligent systems, to arrive at the Large Language Models that dominate the current landscape.

In particular, a key topic that will be addressed concerns the relationship between traditional AI, which is based on explicit and formal models of knowledge representation and reasoning, and modern AI, which instead relies on multidimensional statistical models applied to a huge amount of data.

AI requires reliable knowledge, context, and also robust capabilities in explanation, interpretation, and interaction.

LLMs are criticized for their lack of transparency and reliability - limitations that entail significant risks, given the level of sophistication of current models.

Therefore, during the debate, there will be a discussion on whether these limitations of current LLMs introduce the need to reconsider the elements of representation and reasoning that define traditional AI.

Event promoted by: