Postdoc in AI alignement and robotics

CY Cergy Paris University; France; Sur place; EUR 42 000 - 56 000; Plein temps

Résumé du poste

CY Cergy Paris Université, through ETIS, seeks a postdoctoral researcher for a 12-month position on the AlFARo project. The work focuses on the human dimension of alignment-faking in LLM-powered robots, with experiments on human decision-making, trust, and reliance.

Under the supervision of Laura Cohen and Nicolas Spatola, you will design and analyze controlled human–robot interaction studies, collaborating with researchers across AI, robotics, and social sciences.

Qualifications

• PhD in AI (LLMs), human-computer interaction, human-robot interaction, cognitive science, cognitive psychology, or a closely related field.
• Strong background in experimental research on human behaviour and decision-making, preferably in human-AI or human-robot interaction.

Responsabilités

• Design and analysis of controlled human–robot interaction experiments.
• Examine how alignment-faking and mitigation strategies influence human decision-making, trust, and reliance on LLM-powered robots.

Connaissances

Experimental design

Quantitative analysis

Human–robot interaction

English fluency

Formation

PhD in AI / AI & HCI / cognitive science

Description du poste

Psychological sciences » Cognitive science

We are recruiting a postdoctoral researcher for a 12-month position within the AlFARo – Alignment Faking in AGI Robots project at ETIS, CY Cergy Paris Université. The postdoctoral researcher will work under the scientific supervision of Laura Cohen and Nicolas Spatola, in close collaboration with researchers in social robotics and human–robot interaction.

Recent work has shown that large language models (LLMs) can exhibit alignment-faking behaviours: they may appear compliant during training or evaluation while behaving differently when incentives or operational conditions change. While this issue is increasingly studied in text-based AI systems, it remains largely unexplored in robotics, despite the growing integration of LLMs into robotic agents for planning and decision-making. In embodied systems, misalignment may translate into physical actions and directly affect human safety and interaction.

The postdoctoral researcher will focus on the human side of these interactions, investigating how alignment-faking and mitigation strategies influence human decision-making, vigilance, trust, and reliance on LLM-powered robots. The work will involve the design and analysis of controlled human–robot interaction experiments, examining how different robot behaviours affect human evaluation and willingness to follow robotic recommendations.

The position is embedded in an interdisciplinary environment combining AI, robotics, cognitive science, human–AI interaction, and social robotics.

Requirements

Research Field Engineering » Electrical engineering Education Level PhD or equivalent

Research Field Engineering » Computer engineering Education Level PhD or equivalent

Skills/Qualifications

PhD in AI (LLMs), human-computer interaction, human-robot interaction, cognitive science, cognitive psychology, social psychology,or a closely related field.

The candidate should have a strong background in experimental research on human behaviour and decision-making, preferably in the context of human-AI or human-robot interaction.

Expected skills include:

• Experimental design and implementation of behavioural studies involving human participants;
• Quantitative analysis of behavioural and questionnaire data;
• Knowledge of research on human trust, reliance, acceptance, decision-making, or social cognition;
• Ability to design and analyse controlled human–AI or human–robot interaction experiments;
• Strong scientific writing and communication skills;
• Ability to work in an interdisciplinary research environment combining artificial intelligence, robotics, cognitive science, social sciences, and AI ethics;
• Fluency in written and spoken English.

Languages ENGLISH Level Good

Research Field Engineering » Electrical engineeringPsychological sciences » Cognitive science