Research Area

My primary research area is the large topical intersection of natural language inference (NLI) and inferentialism. NLI is a subdomain of natural language processing (NLP), dedicated to the task of correctly identifying logical relations of entailment and contradiction between claims. Currently, virtually all competitive NLI approaches rely on transformer-based LLMs. However, we hypothesize that LLMs, being world-class associators (system 1 champions, to borrow from John Bargh), might not be ideally suited for logical reasoning, calling for a hybrid, neuro-symbolic approach, combining the strengths of LLMs (system 1) with rule-based methods (system 2). Inferentialism is a school of thought in theoretical philosophy that holds that logical entailment is at the heart of linguistic meaning, understanding, and knowledge; as a consequence, inferentialists have produced a wealth of theoretical insights and proposals on the topic. In my research, I connect these two different disciplines, contributing to both top-tier philosophy journals and A* conferences in NLP. My secondary research area concerns reflection about the kind of things that LLMs are, and on the ethical implications that might ensue from this: Are they tools or some kind of non-biological agents?

Academic Career

I studied (in chronological order) mathematics at ETHZ, linguistics, philosophy, and Multi-Lingual Text Analysis at UZH, where I graduated with an MA in Philosophy and Multi-Lingual Text Analysis. My PhD thesis (summa cum laude, also at UZH) focused on the philosophy of mind and the empiricist inferentialism of W.V.O. Quine. After that, I spent several years working in university management and as a part-time postdoc in Data Science and NLP at the University of St.Gallen. Now I am honored to be PI at project AI-R. Throughout my academic career, I have successfully acquired competitive third-party funding from various institutions, including the SNSF.
I have been teaching on a university level since 2013, predominantly in logics and theoretical philosophy, at UZH and the University of St.Gallen (HSG).

Project

Human belief systems are made up of networks of claims that are connected by logical relationships of entailment and contradiction – the so-called space of reasons. Recent developments in natural language processing and the philosophy of language have opened a unique window of opportunity to explore this space of reasons. Project AI-R seizes this opportunity, aiming to develop a novel hybrid method in natural language inference to build a data-grounded representation of this space of reasons. Following a radically open approach and building on aesthetic insights, the project will develop an interface to allow the general public to interact with (and thereby keep updated) our representation of the space of reasons.