LLMs can shift people's beliefs.
But most persuasion studies only check beliefs before and after a conversation.
We built PersuasionTrace to measure beliefs turn by turn, so we can study how belief updates actually unfold.
Disturbing anecdotal reports of "AI psychosis" and negative psychological effects have been emerging in the news. But what actually happens during these lengthy delusional "spirals"? In our preprint, we analyze chat logs from 19 users who experienced severe psychological harm🧵👇
Can LLMs use ToM to genuinely persuade you, or do they just use good rhetoric?
In our new preprint, we use the MINDGAMES framework to test this. Surprisingly, LLMs like o3 can be incredibly effective persuaders *without* actually understanding your mental states. 🧵👇
In a paper we’re workshopping this week at @NeurIPSConf, @nedcpr@wes_deng@benhutchinson & I ask: what is the model of societal impacts reflected in efforts to evaluate GenAI systems?
Paper: https://t.co/NxMb2mxTNo
Workshop: https://t.co/7z7ZebPcS5
#NeurIPS2024
1/5
Registrations are now open for the #ALTA2024 conference affiliated with @aclmeeting 🎉
📆 Monday 2nd Dec - Wed 4th Dec 2024
📍 Birch Building Innovation Space,
35 Science Road, @ourANU#Online and in-person options available.
https://t.co/tI7QS8QJtn
Prof @ehovy of @UniMelb will deliver a free public #lecture at #ALTA2024:
"Generative #LLMs: How they work and where they are heading"
📷 5.30 - 6.30pm AEDT, Mon, 2nd Dec
📷 Birch Bldg Innovation Space,
35 Science Rd, @ourANU
https://t.co/TuOVHcKwg8
📢 NEW PUB! @nedcpr, @margaret_eby, and I wrote this last Spring to theorize the disorienting impact ChatGPT has on social institutions & norms. 2 years later, everything feels different, but many aspects of AI development remain the same. Read more here:
https://t.co/UgZ0E1Mokg
@IanArawjo I submitted a short paper to CHI last year. The chairs "strongly encouraged" them, but reviewers were less enthusiastic. There were lots of requests for 'more detail'. Mine was luckily accepted (about a 15% rate), but I'm hesitant to try again.
What does it mean to create datasets responsibly for training Multilingual LLMs? New position paper discussing relevant social, cultural, and ethical considerations.
https://t.co/yQ5ReKbrEV
🐝🧠🤖 Curious about how bees think, why human reason might be social, or the challenges in developing AI that understands our world? Last month, I was lucky to learn about these topics and more @DivIntelligence in Scotland. A few reflections below: https://t.co/jByiosnhYx