Samuel Rhys Cox

In thread below sharing details of 3 studies on: - format used when chatbots reference user utterances - LLMs for diverse creativity - deception & social robotics in eldercare publishing at ACM Human-Agent Interaction (@hai_conference) together with @weitsang and collaborators

@What2DoNext @DrSimiManni @ACMIMX Congratulations Simona! Awesome work and loved the paper! 🎉🎉

Congratulations @DrSimiManni and team! Fabulous piece of work to add to @ACMIMX. @SamuelRhysCox, Franziska and I enjoyed reading the paper very much and look forward to seeing more of the work in the future.

@hai_conference @weitsang We contextualise prior use & belief of robotic deception, and draw to attention the potential for future implications (given the increased capabilities of LLMs). With vision, language & robotic models becoming more capable considering such implications is of growing importance

@hai_conference @weitsang Finally, in discussion paper: "The Use of Deception in Dementia-Care Robots: Should Robots Tell "White Lies" to Limit Emotional Distress?" w/ Grace Cheong we overview some deception techniques that care-givers may use when caring for people with dementia doi.org/10.48550/arXiv…

In thread below sharing details of 3 studies on: - format used when chatbots reference user utterances - LLMs for diverse creativity - deception & social robotics in eldercare publishing at ACM Human-Agent Interaction (@hai_conference) together with @weitsang and collaborators

@AI_Conference @pandeyparul @AnthropicAI The "ask the model World Values Survey (WVS) and Pew Research Center’s Global Attitudes (GAS) multiple-choice survey questions as they were originally written". So there are a few examples in the figure and table here (& can see in Fig. how compares across countries)

@pandeyparul @AnthropicAI What are those values specifically?

Ever wonder how well Large Language Models (#LLMs) represent global perspectives? The recent @AnthropicAI study sheds light on some fascinating findings. • It turns out that LLMs seem to echo the views of specific groups, particularly those from the USA and certain European and South American countries. This observation suggests potential biases lurking within these models. • When we ask the model to consider a viewpoint from a specific country, it adjusts its responses accordingly. But sometimes, this adjustment can result in reflecting not-so-positive cultural stereotypes. Here are some key insights from the paper: 1️⃣ There's a real need to evaluate these models more rigorously to understand the values they're echoing. 2️⃣ If we can align these models with a broader range of global perspectives and peel back the layers of the values reflected by AI systems, we'll be taking steps toward tackling social biases. The goal? Models that are truly representative of our diverse world. 🔗 Paper: lnkd.in/ggBvetRN

@panxueni @ismarconf Nice meeting you and hearing about your work, and safe flight!

On my way back to London… I wish I can stay here for another month for great food, amazing natural, and wonderful discussions with fellow researchers about the future of #VR, #MR, #AR, & virtual humans. Thank you @ismarconf for a great conference! #ismar2022 #imsar22

Epic evening with @cwt_sam & @SamuelRhysCox on a boat tour followed by chicken rice & barley tea! We talked about #VR, #AR, memory, cognition, conversation agents, creativity, affective computing, & #NLP #ISMAR2022 #Singapore

The proceedings for CUI 2022 are now available in the ACM digital library @ACM_CUI #cui2022 #chi2023 dl.acm.org/doi/proceeding…

Thank you for the photo @JTrippas! (nice to share with friends and family), and exciting to hear about everyone's work at @ACM_CUI and engage with an exciting community!

Preparing for the exciting discussions to be had at #cui2022? Videos for the programme are up with playlists for the various tracks! ▶️ youtube.com/channel/UCQTUT…

Once again the annual reading list on Conversational User Interface studies at #CHI2022. The list is based on a search with the relevant keywords in the proceedings. Let me know if I missed any. 84 exciting papers on diverse topics! #chatbots #siri #alexa docs.google.com/document/d/1qD…

@AleLuzUXR Reading this made me feel better about not wanting to pay the registration fee (out of my own pocket) so thanks for that though!

TLDR of #CHI2022 experience for the remote participants: “Can you hear me?”; “Can you see the screen?”; “Can you…?”; *talks over one another*

@What2DoNext @BBCRD @Samsung Good luck in your new position!

After 14.5 years with @BBCRD, I am leaving the organisation and moving on to @Samsung this May.

Excited to share that our full paper "Does Chatbot Language Formality Affect Users’ Self-Disclosure?" 🤖🤖 is accepted to #CUI2022 @CUI_Conference Looking forward to presenting our work, and thank you to @weitsang for their amazing supervision!

@SarahBeth721 (Also) also, good luck! (Couldn't fit in first Tweet because I went on too much 👀)

@SarahBeth721 Sounds interesting. Sadly I got this message when clicking "upload image or video". Also not sure if there would be more instructions for writing the description on the upload page (?) but was going to follow the structure of the two examples on your site

Hi Friends! As you're out and about the next several weeks please help me out with some of my PhD research by snapping a quick picture and uploading it to this site: rootedplace.org/data.html Thanks!! #AcademicTwitter #AcademicChatter #Dissertation