Mario Giulianelli
@glnmario
Language and AI Science | Senior research scientist at @AISafetyInst and incoming associate professor @ucl | Previously @ETH_en @illc_amsterdam
Some personal news ✨ In September, I’m joining @ucl as Associate Professor of Computational Linguistics. I’ll be building a lab, directing the MSc programme, and continuing research at the intersection of language, cognition, and AI. 🧵
👾 Full-time research assistant position (1 year) with @sebschu and me! 👾 We're looking for someone to join the research agent evaluation team, starting Fall 2025. Application link to be available soon, but feel free to send us your CV and/or come talk to us at #ACL2025. 🧵
Yes to modelling raw comprehension data! 👀🧠 Huge information loss when using common aggregates, and support for various theories might actually change... Go find @francignare at #ACL2025NLP for juicy details on this spatio-temporal point process model of scanpaths
We present a probabilistic model that jointly generates the three components of each fixation: onset time, landing position (saccade planning) , and duration (duration modelling). Here below an animation of our model on an unseen scanpath sequence (slower version in our repo):
💥 New #ACL2025 main paper!💥 Most #eye #tracking studies collapse fixations into a single “gaze time” metric. We model the full scanpath with a generative probabilistic model. Why? To test how disaggregated modeling stacks up against aggregated approaches.🧵 #Psycholinguistics
📢 I'm looking for a postdoc to join my lab at NYU! Come work with me on a principled, theory-driven approach to studying language, learning, and reasoning, in humans and AI agents. Apply here: apply.interfolio.com/170656 And come chat with me at #CogSci2025 if interested!
please apply for this: civilservicejobs.service.gov.uk/csr/jobs.cgi?j… it's a strategy and delivery manager role (non-technical) in our "AI and Human Influence" team at AISI. Would suit someone who cares about AI policy, and wants to work in a fast-paced environment in the shadow of Big Ben
Omg I love this
Our tool (semantic-features) takes the idea from a method developed by Chronis et al., 2023, and generalizes it to be applied to most Masked Language Models on huggingface. You can use it to project your fav. MLM’s representations to three different semantic-norms.
The goal for this was to advance the science of AI agent evaluations and build common best practices for testing AI agents — what a productive collaboration and what an honour to lead the analysis for the cyber agents strand. I look forward to the next joint testing exercise!
🌏Our largest international joint testing exercise to date brought together participants from across the globe to advance the science of AI agent evaluations. Read more at the link below ⬇️ aisi.gov.uk/work/internati…
To understand what risks AI systems pose, we need to evaluate the upper limits of their capabilities. Post-training enhancements can raise this limit. That's why we’re prioritising elicitation – uncovering the full range of what models can do 🧵