The ELLIS, AI4media, and AIDA Symposium on Large Language and Foundation Models call for deeper interactions and larger-scale initiatives

The arrival of ChatGPT and GPT4 last year marks a new era of AI research on foundation models. While scale is always a key factor in learning, these models have shown that scale has an unexpectedly pivotal impact on generalization and the importance of multimodality. This paves the way for AI subfields, such as language, vision, audio, to converge, not simply to improve numbers by an iota, but as instrumental learning components to generalist AI.

For this crucial reason, the first AI symposium on LLM/Foundation Models focusing on cross-discipline interaction was initiated several months ago. The event was successfully convened on 17-18 October, 2023 in Amsterdam, the Netherlands, under the umbrella of ELLIS, AIDA, and AI4Media, with a view of drawing in a diverse AI population on the topic.

The ELLIS, AI4media, and AIDA Symposium on Large Language and Foundation Models brought together a notable lineup of speakers, panelists, and organisers from different AI subdisciplines in Europe to one of the most advanced technology capitals, Amsterdam.

More than 140 online and in-person participants attended the event, including researchers from various disciplines within AI. The event comprises thematical talks, panel and roundtable discussions, a poster session, as well as a social debate.

The thought-provoking talks which were delivered by renowned AI researchers covered a broad range of perspectives, from language, computer vision, and information retrieval to education and policy. The constructive exchanges on the LLM/Foundation Models further continued at the poster session showcasing 25 state-of-the-art research on the highly-discussed topic. The symposium also facilitated interactive roundtable and panel discussions which attempted to address the most recent challenges on machine teaching, as well as AI and education. The audiences were also engaged in a social debate over imperative measures to balance the development of AI with its societal impact. The critical discussions on the two intensive yet engaging days concluded on a high note, calling for deeper interactions and a larger scale of initiatives on LLM/Foundation Models.

The symposium also underlines the significance of collaboration in tackling emerging topics on AI with AI4Media, AIDA, and the ELLIS Society network joining forces.

In case you missed the symposium, the recording is available.

The symposium was sponsored through the ELLIS unit Amsterdam’s Workshop Support Programme, a programme dedicated for local events initiated and organised by the unit’s ELLIS faculty with the aim to further promote advancement in AI. The programme’s details can be found here.