Format: Workshop · Required preknowledge: Intermediate · Room: EG01 (WS)
Time: 10:45–11:15
Topics: AI Governance
For decades, computing separated the passive data from the rules it follows to process that data. This led us to expect a clear division of labour between a the program sets the rules, and the data is simply what gets read or changed. GenAI break that expectation, and now the text we enter, the document we provide, the sources we include no longer sit still as material to be analysed. They shape the very rules by which the model analyses them, and in doing so influence every part of the outcome.
For any profession built on reading, understanding, and interpreting texts, understanding this shift is fundamental and a purely technical account does little to explain what it means in practice. To grasp the impact, we must look at language itself: how language works, how language models work with it, and how we in turn work with both. This workshop takes apart the vulnerabilities that open up when human understanding of meaning and the machine’s processing of tokens diverge, and helps professionals recognise and manage the risks that arise when the text meant to be analysed rewrites the rules of its own analysis.
Speakers: Marc Heimann