Lennart Wachowiak

Lennart Wachowiak

Postdoctoral Researcher at University of Oxford | Technical AI Safety Fellow at Cambridge ERA:AI

University of Oxford

Biography

I am a Cambridge ERA:AI research fellow working on human–AI interaction and decision-making, and an incoming postdoctoral researcher at the University of Oxford. I received my PhD from the CDT in Safe and Trusted AI at King’s College London and Imperial College London. Previously, I worked as a student researcher at Google DeepMind, where I was part of the Gemini Robotics team.

Decision-Making. My current research focuses on how AI systems can help people make better decisions, especially in settings where other AIs may try to manipulate users into harmful choices. In particular, I study how an advisory AI can detect and counter manipulative or adversarial influence.

Explainability. During my PhD, I investigated how robots can learn to proactively explain their decisions and assist people in collaborative scenarios. I trained machine learning models to predict when explanations are needed and what they should be about by interpreting the user’s non-verbal behaviour and contextual task cues.

Cognitive Linguistics. I love interdisciplinary work, learning about different perspectives, and combining them to get a rich understanding of a phenomenon. Beyond my work in human–AI interaction, I study conceptual metaphors and image schemas in text corpora using tools from natural language processing and theories from cognitive linguistics.

Download my CV.

Work

  • Postdoctoral Researcher, Incoming

    University of Oxford

  • Technical AI Safety Fellow, 2026

    Cambridge ERA:AI

  • Student Researcher, 2024-2025

    Google DeepMind

Education

  • PhD in Safe and Trusted AI, 2021-2026

    King's College London, Imperial College London

  • MSc in Cognitive Science, 2018-2021

    Joint Degree with University of Vienna, University of Ljubljana, Comenius University Bratislava, Eötvös Loránd University Budapest

  • BSc in Computer Science, 2014-2017

    University of Paderborn

Research Areas

Explore the research areas I have investigated.

Human–AI Decision-Making
I investigate how (i) AI can prevent people from being manipulated into suboptimal decisions, and (ii) how AI can facilitate finding common ground ingroup-decision-making.
Human–AI Decision-Making
Explainable Robotics
My PhD thesis focuses on explainable robotics. I develop user-centered and proactive explanations by allowing robots to consider a user’s mental state, inferred from social cues and task context.
Explainable Robotics
Conceptual Metaphors and Image Schemas
How can we identify and analyse conceptual metaphors and image schemas in natural language texts? In what contexts do people use them? What words and syntactic structures indicate them?
Conceptual Metaphors and Image Schemas
Extracting Terminological Concept Systems
How can we extract and organize domain-specific knowledge from multilingual text?
Extracting Terminological Concept Systems

Publications

Quickly discover relevant content by filtering publications.
Lennart Wachowiak, Scott Blain, David Williams-King, Samuele Marro (2026). LLM Wardens: Mitigating Adversarial Persuasion with Third-Party Conversational Oversight. Preprint.

PDF Cite Code Project

Lennart Wachowiak, Andrew Coles, Gerard Canal, Oya Celiktutan (2026). What Questions Should Robots Be Able to Answer? A Dataset of User Questions for Explainable Robotics. Preprint.

PDF Cite Code Dataset Project

Lennart Wachowiak, Andrew Coles, Oya Celiktutan, Gerard Canal (2026). Neurosymbolic Explanation Selection in Robotics: Combining the Strengths of Planning and Foundation Models for XAI. HRI Compendium.

PDF Cite Code Project DOI

Lennart Wachowiak, Peter Tisnikar, Andrew Coles, Gerard Canal, Matteo Leonetti, Oya Celiktutan (2025). Predicting When and What to Explain From Multimodal Eye Tracking and Task Signals. Transactions on Affective Computing.

PDF Cite Code Project DOI

Lennart Wachowiak, Peter Tisnikar, Andrew Coles, Gerard Canal, Oya Celiktutan (2024). A Time Series Classification Pipeline for Detecting Interaction Ruptures in HRI Based on User Reactions. ICMI.

PDF Cite Code Project DOI

Lennart Wachowiak, Andrew Coles, Oya Celiktutan, Gerard Canal (2024). Are Large Language Models Aligned with People's Social Intuitions for Human-Robot Interactions?. IROS.

PDF Cite Code Project DOI

Lennart Wachowiak, Andrew Coles, Gerard Canal, Oya Celiktutan (2024). A Taxonomy of Explanation Types and Need Indicators in Human–Agent Collaborations. Journal of Social Robotics.

PDF Cite Dataset Project DOI

Philipp Wicke, Lennart Wachowiak (2024). Exploring Spatial Schema Intuitions in Large Language and Vision Models. ACL Findings.

PDF Cite Code Project DOI

Lennart Wachowiak, Andrew Fenn, Haris Kamran, Oya Celiktutan, Andrew Coles, Gerard Canal (2024). When Do People Want an Explanation from a Robot?. HRI.

PDF Cite Code Project DOI

Lennart Wachowiak, Dagmar Gromann, Chao Xu (2023). The Image Schema VERTICALITY: Definitions and Annotation Guidelines. Image Schema Day.

PDF Cite Project

Lennart Wachowiak, Dagmar Gromann (2023). Does GPT-3 Grasp Metaphors? Identifying Metaphor Mappings with Generative Language Models. ACL.

PDF Cite Code Project Video

Lennart Wachowiak, Oya Celiktutan, Andrew Coles, Gerard Canal (2023). A Survey of Evaluation Methods and Metrics for Explanations in Human–Robot Interaction (HRI). ICRA Explainable Robotics Workshop.

PDF Cite Project

Lennart Wachowiak, Dagmar Gromann, Chao Xu (2022). Drum Up SUPPORT: Systematic Analysis of Image-Schematic Conceptual Metaphors. EMNLP Workshop FLP.

PDF Cite Code Project Video DOI

Lennart Wachowiak, Dagmar Gromann (2022). Systematic Analysis of Image Schemas in Natural Language through Explainable Multilingual Neural Language Processing. COLING.

PDF Cite Code Project

Lennart Wachowiak, Peter Tisnikar, Gerard Canal, Andrew Coles, Matteo Leonetti, Oya Celiktutan (2022). Analysing Eye Gaze Patterns during Confusion and Errors in Human–Agent Collaborations. RO-MAN.

PDF Cite Code Project Video DOI

Peter Tisnikar, Lennart Wachowiak, Gerard Canal, Andrew Coles, Matteo Leonetti, Oya Celiktutan (2022). Towards Autonomous Collaborative Robots that Adapt and Explain. ICRA Workshop on Prediction and Anticipation Reasoning in HRI.

PDF Cite Project Poster

Dagmar Gromann, Lennart Wachowiak, Christian Lang, Barbara Heinisch (2022). Extracting Terminological Concept Systems from Natural Language Text. Cognitive Technologies.

PDF Cite Project

Christian Lang, Lennart Wachowiak, Barbara Heinisch, Dagmar Gromann (2021). CogALex 2.0: Impact of Data Quality on Lexical-Semantic Relation Prediction. NeurIPS DCAI Workshop.

PDF Cite Code Dataset Project Video

Dagmar Gromann, Christian Lang, Lennart Wachowiak, Barbara Heinisch (2021). Multilingual Extraction of Terminological Concept Systems. Workshop on Deep Learning and Neural Approaches for Linguistic Data.

PDF Cite Project

Lennart Wachowiak, Christian Lang, Barbara Heinisch, Dagmar Gromann (2021). Towards Learning Terminological Concept Systems from Multilingual Natural Language Text. Conference on Language, Data and Knowledge (awarded best student paper).

PDF Cite Code Project Video DOI

Christian Lang, Lennart Wachowiak, Barbara Heinisch, Dagmar Gromann (2021). Transforming Term Extraction: Transformer-Based Approaches to Multilingual Term Extraction Across Domains. ACL Findings.

PDF Cite Code Project Video DOI

Lennart Wachowiak, Christian Lang, Barbara Heinisch, Dagmar Gromann (2020). CogALex-VI Shared Task: Transrelation - A Robust Multilingual Language Model for Multilingual Relation Identification. Workshop on the Cognitive Aspects of the Lexicon (1st place in CogALex shared task).

PDF Cite Code Project

Lennart Wachowiak (2020). Semi-automatic Extraction of Image Schemas from Natural Language. MEi:CogSci Conference.

PDF Cite Project

Rebecca Varga, Lennart Wachowiak (2019). Grounded Language Learning in Robots — An Information-Theoretic Approach to Adjective Learning. MEi:CogSci Conference.

PDF Cite Project

Contact