Lucio La Cava

🧠 🕵️ LLMs are mastering the "imitation game"... but can we still tell who wrote what?

📝 At #EMNLP2025, we present OpenTuringBench — 500K+ texts, 7 open LLMs, 7 challenging tasks, and a new method for machine-generated text detection and attribution!

📍 Today at 2 PM in Hall C

Thrilled to be visiting the NEtwoRks, Data, and Society (@nerdsitu.bsky.social) group at @itu.dk!

After a wonderful experience here during my PhD, it’s a pleasure to continue collaborating with @lajello.bsky.social 👨🏻‍💻

Excited for another inspiring chapter in Copenhagen! 🇩🇰

🤖 Do LLMs reflect our moral expressions—or alter them?

In our #ACL2025 paper, we study 12 widely used LLMs and found that moral expressions are altered to varying degrees depending on the editing task and moral conditioning prompt.

📍 ACL2025 — Hall X5 (28)
🕚 Today, 11:00–12:30

Thank you, Matteo, for the warm hospitality during our visit to Uppsala University! It was a pleasure teaching a Large Language Model course to such passionate students. I truly appreciated the inspiring exchange of ideas with the UU-InfoLab members, and I look forward to future collaborations!

This week we had the pleasure to host our colleagues and friends from Università della Calabria. @luciolcw.bsky.social delivered a well-attended intensive course on Large Language Models, and Andrea Tagarelli presented the research done by his team. We look forward to new collaborations.

📝 We reveal varying levels of alteration of moral expressions depending on the type of text modification task and moral-related conditioning prompt.

Looking forward to presenting our work at #ACL2025 in Vienna! 🇦🇹🥳

🤖 Our study analyzes the behavior of 12 LLMs among the most widely used Open and uncensored (i.e., "abliterated") models over human-annotated datasets.

🔍 Using Moral Foundation Theory, we investigate how Large Language Models (LLMs) capture and potentially alter moral dimensions through text manipulation.

🎉 Excited to share that our paper "Exploring LLMs' Ability to Spontaneously and Conditionally Modify Moral Expressions through Text Manipulation" has been accepted to #ACL2025 (Main)!

Yesterday, we had the pleasure of e-meeting the OpenAI team for an insightful discussion about Open LLMs. While I can’t share details just yet, there’s a lot to be excited about! 🚀

It was particularly inspiring to see Sam Altman join the session and share OpenAI’s vision for open models 🤯

🚨 Just 1 week to go! ⏳

Don't miss your chance to submit short papers and 2-page abstracts to the Next-Gen and Alternative Social Media Workshop at @icwsm.bsky.social — deadline is April 14 (AoE)! ⏰

👉🏼 nextgensocial-workshop.github.io

🚨 Deadline extended to April 14 📅

📚 Extra time to submit your abstracts and short papers to the Next-Gen and Alternative Social Media Workshop at @icwsm.bsky.social!

👉🏼 nextgensocial-workshop.github.io

🔥 Are you working on next-gen or alternative social media platforms like #Bluesky or #Mastodon?

👇🏻 This @icwsm.bsky.social workshop is for you! 😎
📢 nextgensocial-workshop.github.io

📚 We solicit research, position, demo, and dataset paper submissions!
📆 Submissions: March 31, 2025

Wrapping up an amazing #AAAI2025 in Philadelphia 🇺🇸

We presented our work "Open Models, Closed Minds? On Agents Capabilities in Mimicking Human Personalities through Open Large Language Models"

📝 arxiv.org/abs/2401.07115

Grateful to everyone who engaged with our work for the valuable feedback!

🔥 Are you working on next-gen or alternative social media platforms like #Bluesky or #Mastodon?

👇🏻 This @icwsm.bsky.social workshop is for you! 😎
📢 nextgensocial-workshop.github.io

📚 We solicit research, position, demo, and dataset paper submissions!
📆 Submissions: March 31, 2025

🔥 Hey NetPals!

This week, we welcome Lucio La Cava @luciolcw.bsky.social from University of Calabria, who will discuss the rise of decentralized social media and challenges for early-career researchers in the age of big tech.

📅 Mark your calendars: 6th Feb, 4pm UK time

Zoom: netplace.site

Had an amazing time last week in Trento for #CS2Italy! 🇮🇹

I presented our work on automating content compliance checking in social media using LLM agents (arxiv.org/abs/2409.08963) 🤖📝

It was great to catch up with old friends and meet new ones! So grateful for the inspiring conversations 🙌

group photo of 12 happy people in front of orange and blue colored windows

Huge turnout of NERDS (+alumni) at #cs2italy ! It is so great to get together + meet up with former visitors/group members.

Open Models, Closed Minds? On Agents Capabilities in Mimicking Human Personalities through Open Large Language Models The emergence of unveiling human-like behaviors in Large Language Models (LLMs) has led to a closer connection between NLP and human psychology. Scholars have been studying the inherent personalities ...

🎉 Exciting news! Our paper "Open Models, Closed Minds? On Agents Capabilities in Mimicking Human Personalities through Open Large Language Models" has been accepted at the 39th Annual AAAI Conference on Artificial Intelligence #AAAI2025! 🥳

📝 Check it out here: arxiv.org/abs/2401.07115

#NLP

I'd be happy to be added!

Would like to be added! 🙋🏻‍♂️

Ogni tanto capita di dover andare in giro 🙆🏻‍♂️

Wrapping up an amazing #EMNLP2024 in Miami! 🌴☀️

Grateful to everyone who visited our poster for their valuable feedback and stimulating discussions! 🤗

I’d like to be included! 🙋🏻‍♂️

Would love to be included, thanks!

I’d like to join 🙋🏻‍♂️

🌴 Excited to head to #EMNLP2024 next week! I'll be presenting our work "Talking the Talk Does Not Entail Walking the Walk: On the Limits of Large Language Models in Lexical Entailment Recognition" 📝

📄 Check out the preprint here: arxiv.org/abs/2406.14894

👨🏻‍💻 Glad to have presented our work "Is Contrasting All You Need? Contrastive Learning for the Detection and Attribution of AI-generated Text" 🤖 at ECAI 2024 in Santiago de Compostela 🇪🇸

📚 doi.org/10.3233/FAIA...

We also walked the final section of the Santiago’s French Way! ✨

📢 Happy to share that our paper "Talking the Talk Does Not Entail Walking the Walk: On the Limits of Large Language Models in Lexical Entailment Recognition" has been accepted at EMNLP'24 Findings #EMNLP2024 🥳☀️🌴

📝 Paper: arxiv.org/abs/2406.14894

Really enjoyed presenting our tutorial on “Mining, Modeling and Analysing Decentralised Social Media” w/ Alessia Galdeman during the #ASONAM24 Conference 🥳

My focus was on working with free and open APIs from Decentralised Social Platforms in an API-scarcity era 👨🏻‍💻

#Mastodon #Fediverse #Bluesky