alt.britchi: Strange Relations with AI
Date: Monday 2 Nov 2026
Time: 12:00–12:30
Add this session to your calendar, or click a talk’s time to add just that talk.
| Time | Presentation | Abstract | Authors | Contribution type |
|---|---|---|---|---|
| 12:00-12:10 | AI is a Gungfly: Why We Need New Mental Models for LLMs | Large Language Models (LLMs) are now used for a large area of applications. However, everyday users need better mental models to understand and navigate these Artificial Intelligence (AI) systems. In HCI, mental models are historically shaped by deterministic computer systems delivering consistent outputs. LLMs overturn this principle, delivering what are essentially probabilistic outputs. This can be a strength, but the variability of outputs and notably the untruths can cause misalignment between deterministic user expectations and probabilistic system functionality. Because the inner workings of AI systems are hard to determine, we used synthetic ethnography to observe user inputs and system outputs. Using a synthetic probe, we observed variable iterative outputs from a simple calendar-based question. Based upon this, we suggest users accept the data landscape of LLMs is not reliable terra firma, but more like a gungfly – a Swedish term for a place where the ground is unstable and changeable. | Hunt, Nemeth, Holmquist | alt.britchi |
| 12:10-12:20 | Talking Shit with ChatGPT | Chatbots powered by large language models (LLMs) like ChatGPT are well-known bullshitters. In some sense, this is an expected function of how generative artificial intelligence (AI) works and the garbage that feeds them, i.e., all sorts of junk from around the Internet. Dominant narratives have zeroed in on the problem of hallucinations. In this provocation, we aim to shift the conversation towards greater nuance (while retaining the fecal verbiage). With ChatGPT as our scapegoat, we explore the notion of shit-talking within conversational interactions. We raise theoretical questions, propose harmful and potentially worthwhile applications, and discuss design strategies. We include demonstrations with ChatGPT, for good measure. Gather ’round and join us in talking shit about—and with—ChatGPT. Content Warning: AI-generated insults, personal attacks, and derogatory speech. | Seaborn, Desai | alt.britchi |
| 12:20-12:30 | It’s Time to Think About Time: Artificial Intelligence, the Speed of Organisational Culture, and the Social Acceleration of Knowledge Work | Not only is Artificial Intelligence changing the pace of work and culture, but also the pace of change itself. To the people living through it, this can feel like whiplash. To describe that change, we typically reach for languages of technological progress, economic growth, labour process, or occupational pressures. This paper introduces a different language from the philosophies of time, speed, and social acceleration. For more than a century, philosophers of speed have asked what speed does to culture, to power, and to the self, but their thinking has barely been brought to bear on the changes we are experiencing due to generative AI. We start from the position that AI can be read as a chronotechnology: a time machine that closes the gap between wanting a piece of work and having it, and that produces many futures at once. We then ask how AI acceleration reshapes the knowledge worker’s relationship to the past, the present, the future, to power, and to risk. Our aim is to open the conversation between the discourse on AI and the philosophy of time, a conversation that we believe is currently missing, yet important and productive. | Sarkar | alt.britchi |