Feed them slop: Alignment, attunement, and the affective lure of slopaganda

From Charlie Kirk reborn as Terminator to bearded ladies dancing at cocktail parties in ‘Trump Gaza’, social media feeds now churn with slopaganda or ‘agitslop’. Unlike deepfakes, which lure viewers into mistaking AI-generated videos for reality, slopaganda is obviously fake but affectively real. As the related term ‘agitslop’ suggests, slopaganda carries an affective lure. If it manages to hook a viewer, it stirs up emotional sediment – insecurities, indignations, and petty grievances. As many have noted, generative AI models trained on vast troves of internet comments and social media posts inevitably reflect and refract deep-seated human biases. But the affective resonance of AI slop goes further, both technically and socially. Drawing on the concept of ambient rhetoric, which describes rhetoric not as persuasion between individual agents but as a process of attunement to conditions, moods, and affordances, this article frames AI slop as a productive resonance between ambient rhetoric and supervised fine-tuning and reinforcement learning through human feedback. Beyond inheriting biases from training data, fine-tuning and alignment through reinforcement learning through human feedback train models to ‘feel out’ which responses resonate in sociotechnical environments. In this way, AI models are optimised for rhetorical attunement. Slop also responds to the insatiable appetite of the social media ‘feed’. In this sense, AI slop is not merely an error or excess but an ambient rhetoric highly attuned to dominant norms and technical incentives. When AI slop intervenes in politics as slopaganda or agitslop, it reflects how trained, fine-tuned, and aligned AI models fold into and intensify existing ideological, rhetorical, and technical formations rather than disrupting them. Attending to the affective resonance of slopaganda thus offers a way to reckon with AI systems as ideological attunement machines that optimise feeling, relevance, and response within sociotechnical environments.