Invitation to the online conference: “How to evaluate the quality of behavior of a genIA model?”
Invitation to the online conference with Prof. Luca Mari on the topic entitled: “How to evaluate the quality of behavior of a GenIA model?”
Category: AI and LLM models
Dostoevsky’s Artificial Intelligence, Reflections on the Future – Video Conference
The event emphasizes how interacting with these technological entities requires a humanistic approach, suggesting that human skills that cannot be replaced are related to responsibility and ethics, not just cognitive abilities.
Aiaf Scientific Committee Meeting – GenIA – Video Conference
Dialogue between Engineer Alfonso Scarano and Professor Luca Mari, an expert in generative artificial intelligence, focusing on the importance and challenges of this technology.
Gen AI – Round Table
Cultural and cognitive transformation brought about by generative artificial intelligence, emphasizing how this revolution is not only technical but also social. Schools, central places for learning, need to prepare for these challenges, going beyond just transferring information.
RAG techniques for GenIA – Video Conference
Retrieval-Augmented Generation (RAG) is a technology for improving the responses of artificial intelligence models, it works by transforming questions into numerical vectors through a technique called embedding, and then retrieving relevant documents from a vector database.
The ethical dimension of LLMs – Video Conference
Traditional software is entirely programmable, a generative AI system has its own learning capability that makes it complex to regulate. This raises questions about who should be responsible for the ethical training of these systems and how to prevent inappropriate uses, a critical issue given their potential impact on society and culture.
“Good teachers” to vertically specialize LLM models – Video Conference
Video conference promoted by CERSTE in which Professor Luca Mari delves into the operation of neural networks and emphasizes the distinction between training and network operation, explaining that, unlike traditional software, a neural network must be trained with two main phases.