Social Science One Insight SFI Research Centre BigData@Heart My Health Record Now that academic institutions are increasingly pushed towards cooperating with other industries and forming partnerships, what is their role and position with regard to data ethics and governance? arxiv.org/abs/2301.12347

How do LLMs operationalise 'truth' and how might we frame an anatomy of truth for OpenAI's language models? Luke Munn Liam Magee and I did a bit of an exploration of what kinds of truth are desired and constructed for and by these systems. #chatGPT #LLM #InstructGPT 👇

ChatGPT is producing excitement and panic, part of a general wave of automation reaching into creative labour and knowledge work. Gus Gollings and I discuss implications of its persuasive techniques for management and leadership. #ChatGPT automatedsubjects.medium.com/charismatic-au…

The Legacy Mural Initiative, a public mural art project, aims to address social issues in Tafelsig, a township located in Mitchell’s Plain, Cape Town. Read more about the healing power of art here: use.metropolis.org/newsroom/revit…

Tomorrow: 📢 #TUMI2023 Amplifying feminist voices in transport to drive equitable and sustainable mobility. Let's pave the way for a future where transportation leaves no one behind. ✨🚶‍♀️🚴‍♂️🌍 #FeministTransport #inclusivemobility #WomenMobilizeWomen #TUMI2023

Sustainable transport is key to a greener future for our 🌍. Jakki Mann's essay "Accelerating Change: The Journey towards Sustainable Mobility" highlights the importance of knowledge exchange & collaboration. LINK👉bit.ly/3IERYZB #SustainableMobility #thereisnoplanetB

"Truth Machines" is now published in AI & Society. The paper explores how large language models like ChatGPT reconfigure knowledge claims in important ways. link.springer.com/article/10.100… Our earlier thread on the preprint 👇 steps through the key ideas and arguments.

We are so privileged to be joining everyone today to launch Foundations for Belonging 2023. Kicked off by Madison Shakespeare with a beautiful smoking ceremony. SSI

Wondering what might be at stake for #heritage #visualstudies and the #gaze in the weird new world of #Midjourney #Stablediffusion? Check out this working paper by Liam Magee Luke Munn and I - 'Reframing Built Heritage through the Machinic Gaze' arxiv.org/abs/2310.04628

I'm at 4S talking Generative AI and synthetic monuments at "The politics and aesthetics of synthetic media," killer panel organised by Thao Phan and Fabian Offert , Sat 10:40, 301B, #4SHonolulu23 #genai #photography #gaze

Very excited that this is FINALLY out! When Liam Magee set up #AutomatedSubjects at Institute for Culture and Society (ICS) and recruited Luke Munn and I, we had no clue it would be this fun! This piece brings together critical AI studies with #Lacan and #Freud as one does journals.sagepub.com/doi/10.1177/20…

New Article: Analyzing #AI as 'automated subjects' using #psychoanalysis & #criticalmediastudies. Magee et al.'s study delves into LLMs like #OpenAI's #InstructGPT, exploring their design, biases, and interactions with users. #AILanguage #CriticalAI buff.ly/48YUepj

Check out this piece, hot from the online press! Liam Magee Luke Munn and I attempt to conceptualize to 'machinic gaze' and it's potential implications for how heritage sites might be visualized. #genAI #machinicgaze journals.sagepub.com/doi/10.1177/14…

"The Missing Human in Misinformation Fixes", my new article in Scientific American 🆕🗞️: scientificamerican.com/article/theres… The Centre for Digital Cultures & Societies UQ ADM+S Centre UQ News #disinformation #misinformation

Just #arxived this fun set of #genAI explorations with Liam Magee where we /imagine the #Anthropocene through iterations of an old-school Backus-Naur Form. Builds on some of our #machinicgaze work, but goes into a slightly more speculative space. arxiv.org/pdf/2405.09734

Research conducted by SSI and Western Sydney University strengthens the case for seeing digital inclusion as important as social inclusion in settlement policy and services. Read the full article here: tandfonline.com/doi/full/10.10…

ADM+S researchers have developed a novel design toolkit to help users structure prompts and interrogate bias in outputs of generative AI tools, after discovering LLMs exhibit human-like biases when analysing controversial topics. Learn more ➡️ admscentre.org.au/research-finds…

Liam Magee New research explores the 'problem of alignment' in AI language models, examining historical debates on language structure & statistics. Offers critical perspective on current AI alignment efforts. Read the full paper: link.springer.com/article/10.100… #AIresearch #NLP

Fresh off the press. Our analysis of alignment and its roots in historical debates about linguistic science and statistics, with Liam Magee and Karen Soldatic link.springer.com/article/10.100…

In a recent TechXplore article Liam Magee was highlighted for his crucial role as a mentor in a study conducted by researchers at UQ News. The study focused on how LLMs could enhance human analytical abilities in qualitative research. tinyurl.com/yc5pyzta