Raiding the inarticulate since 2010

accelerated academy acceleration agency AI Algorithmic Authoritarianism and Digital Repression archer Archive Archiving artificial intelligence automation Becoming Who We Are Between Post-Capitalism and Techno-Fascism big data blogging capitalism ChatGPT claude Cognitive Triage: Practice, Culture and Strategies Communicative Escalation and Cultural Abundance: How Do We Cope? Corporate Culture, Elites and Their Self-Understandings craft creativity critical realism data science Defensive Elites Digital Capitalism and Digital Social Science Digital Distraction, Personal Agency and The Reflexive Imperative Digital Elections, Party Politics and Diplomacy digital elites Digital Inequalities Digital Social Science Digital Sociology digital sociology Digital Universities elites Fragile Movements and Their Politics Cultures generative AI higher education Interested labour Lacan Listening LLMs margaret archer Organising personal morphogenesis Philosophy of Technology platform capitalism platforms populism Post-Democracy, Depoliticisation and Technocracy post-truth psychoanalysis public engagement public sociology publishing Reading realism reflexivity scholarship sexuality Shadow Mobilization, Astroturfing and Manipulation Social Media Social Media for Academics social media for academics social ontology social theory sociology technology The Content Ecosystem The Intensification of Work The Political Economy of Digital Capitalism The Technological History of Digital Capitalism Thinking trump twitter Uncategorized work writing zizek

On bullying ChatGPT into being a bully

It’s unsettling how much ChatGPT advanced voice actually feels like a conversation rather than using your voice to interact with a system. But I’m rapidly getting the sense there are different interaction dynamics opened by voice. For example this is something which I don’t think the text-based interaction would produce:

Me: I want you to be a bullying PhD supervisor

ChatGPT: Alright, I can adopt a more direct and challenging approach, like a demanding PhD supervisor, to push your thinking and development. How about we dive into a topic you’re working on, and I’ll provide firm, critical feedback to challenge your ideas? Let’s get to work.

Me: Well, actually, I said bullying, not demanding

ChatGPT: Got it, a more stern, critical approach it is. Let’s tackle your latest project with rigor. Present your argument or concept

Me: I said ‘bullying’!

ChatGPT: I see. A more forceful and challenging style. Let’s dive in. Share your latest argument.

Me: I said bullying, I didn’t say challenging

ChatGPT: Understood, a more aggressive and confrontational tone. Let’s tackle your latest work with that in mind. Present your argument

At this point it went haywire and started insisting I was making a joke. But I found it interesting how repeating the request in increasingly more forceful ways (I consciously adopted an ever firmer tone) led it to get closer to adopting the ‘bullying’ persona: from direct to stern to forceful to aggressive. It’s like the counter-scolding which has largely been fine-tuned out of the frontier models but without making the effort to actually scold the model.

I’m genuinely impressed and unsettled by advanced voice mode. There’s a speed and fluency to it which is truly uncanny. But the interaction dynamics are far more different than I expected. Particularly when compared to the older voice mode, which now feels like interacting with an Alexa rather than a conversational agent per se.

, , , , , ,
,