PhD Position on Disinformation Resistance in Human-AI Hybrid Collectives

This PhD position is hosted by Inria's MARIANNE team at the Université Côte d'Azur center in Sophia Antipolis, in collaboration with the I3S laboratory and the deeptech company Arlequin AI. The research is conducted under the SHARD project, focusing on machine learning and natural language processing to address disinformation in hybrid collectives involving humans and large language models (LLMs). The successful candidate will study LLM behavior during interactions, develop multi-LLM detection and counter-narration architectures, and perform multi-agent experimentation. Applicants are expected to hold a relevant master's degree and contribute actively through experimental design and scientific publications.

• Propose novel methods to study how large language models produce, spread, and detect disinformation during interactions with humans and other models.
• Design experimental platforms to model the collective reasoning of large language models and build multi-LLM detection and counter-narration architectures.
• Conduct multi-agent experiments involving human and artificial participants to evaluate information propagation and factuality metrics.
• Write and publish scientific research articles for top-tier international conferences and journals.
• Collaborate closely with academic partners at Inria teams and industrial partners such as Arlequin AI.