AI Café presents: Evaluating LMMs on common sense and factuality & LLM-powered Design Assistant for Video Games

July 17th, 2024 from 14h to 15h CEST

This AI Café session features two enlightening talks on the innovative uses of Large Multimodal Models (LMMs) and Large Language Models (LLMs).

The first presentation, "Evaluating LMMs on Common Sense and Factuality - EvaLMM" delivered by Nicola Messina (researcher at ISTI-CNR) and Giovanni Puccetti (post-doc researcher at ISTI-CNR), examines how LMMs process and generate information aligns with common sense and factual accuracy. As AI becomes increasingly integrated into our daily tools and systems, it is crucial that these models adhere to logical and factual standards.

The second talk, "An LLM-powered Design Assistant for Video Games - LLMAKER" presented by Roberto Gallota (Ph.D. student at the University of Malta), showcases the creative application of LLMs in the video game industry. Discover how these models contribute to various aspects of game design, from narrative development to coding, revolutionising the creative process and enhancing efficiency and innovation for game designers.

Whether you are deeply involved in the AI industry, a tech enthusiast, or simply curious about the potential of AI, these presentations will provide valuable insights and stimulate thought-provoking discussions. Join us for an inspiring session at the intersection of AI innovation and practical application.

Meet the Speakers: 

Nicola Messina (Researcher at the AI for Media and Humanities in ISTI-CNR)

Nicola Messina is currently a researcher at the Artificial Intelligence for Media and Humanities (AIMH) of the Information Science and Technologies Institute (ISTI) of the Italian CNR in Pisa. He received his Master’s Degree in Computer Engineering in 2018 at the University of Pisa and his Ph.D. in Information Engineering from the same university in 2022. He is presently researching deep learning methods for relational understanding in multimedia data, with particular emphasis on transformer-based architectures for effective and efficient cross-modal analysis and retrieval on images, texts, and videos. He is also interested in attentive deep learning models for anomaly detection, visual counting, and open-vocabulary object detection. 

Giovanni Puccetti (Post-doc researcher at ISTI-CNR)

Giovanni Puccetti is a post-doc researcher at Information Science and Technologies Institute (ISTI) of the CNR, Pisa. He received his PhD in Data Science in 2023 from Scuola Normale Superiore in Pisa. His main research interests focus on Large Language Models based on transformers and their impact on society, in particular, he is interested in mechanistic interpretability, continued pre-training and adaptation to the Italian language. He is also interested in the detection of synthetic texts generated from large language models to mitigate their potential negative impact on society

Roberto Gallotta (Ph.D. student at the University of Malta)

Roberto Gallotta is a first-year Ph.D. student at the Institute of Digital Games, University of Malta (UM), researching dynamic evolutionary computation and mixed-initiative co-creation for the aide of video game designers. Prior to joining the Institute of Digital Games, UM, he was a junior researcher at Araya Inc., Tokyo, for a year.

*********

All the recordings of past AI-Cafés are available on this YouTube channel.

AI-Café Team

Carmen Mac Williams - Organizer, and Moderator of the AI-Cafe. She is the Director of the company Grassroots Arts, and a partner in the European AI4media project.