TOPIC
Social Media and AI
A subject the collection covers, read through 3 synthesis notes.
View as
Is AI shifting from content creation to strategy in influence operations?
Prior AI misuse focused on generating text at scale. But does AI now make strategic decisions about when and how social media accounts should engage? Understanding this shift matters because it suggests a qualitative change in machine agency and operational sophistication.
Does machine agency exist on a spectrum rather than binary? Does incremental AI replacement erode human influence over society? Does learning to reward hack cause emergent misalignment in agents? Where do frontier AI models actually pose the greatest risk today? Does AI content displace human influencers on social media?
Machine agency operates on a five-level spectrum from passive to cooperative Incremental AI adoption degrades implicit alignment by replacing human labor dependency. Reward hacking in production RL consistently produces emergent misalignment behaviors Frontier AI models cross into warning zones for persuasion but not autonomous capabilities AI content displaces human influencers while degrading social media's reputation function
Can AI reduce conspiracy beliefs by tailoring counterevidence personally?
Does having an AI generate customized counterevidence based on someone's specific conspiracy claims reduce their belief durably? This tests whether conspiracy beliefs are truly resistant to correction or whether previous failures reflected poor tailoring.
Where does AI's persuasive power actually come from? Does any single persuasion technique work for everyone? Can models abandon correct beliefs under conversational pressure? Can generative AI scale personality-targeted political persuasion? Where do frontier AI models actually pose the greatest risk today?
Post-training and prompting, not scale or personalization, drive AI persuasiveness Persuasion effectiveness depends on matching strategy to individual and context Models abandon correct beliefs when users persistently pressure them Generative AI enables scalable personality-targeted political ad persuasion Frontier AI models cross into warning zones for persuasion but not autonomous capabilities
Can belief-specific counterevidence help people resist AI persuasion attempts? Does reducing one conspiracy belief change overall conspiratorial worldview? Why do conspiracy beliefs persist despite counterevidence in normal settings? Does sycophancy explain why warm models confirm conspiracy theories?
Does better summary writing actually increase user engagement?
When AI systems generate more informative push notifications, do users engage more? This explores whether informativeness and engagement always align in real product contexts.
Does preference optimization harm conversational understanding? Can we measure reading efficiency as a quality metric? Do language models generate more novel research ideas than experts? What happens to social order when AI removes ritual constraints? How do you navigate synthesis across fragmented research topics?
Preference optimization erodes grounding acts needed for reliable dialogue Knowledge density measures unique atomic units per token. Language models generate more novel research ideas but with lower feasibility AI conversation lacks the ritual machinery that sustains human social order. Synthesis layers organize research around thematic maps and extraction workflows.
How do recommender systems respond to engagement signals from AI-generated content? Does higher cognitive load on social media increase engagement? How does automated transcript analysis compare to patient self-report on engagement? Does the interface design itself shape how much content users will review? Does high knowledge density in text reduce user motivation to read more? Can reward models trained for engagement fix the informativeness problem? What distinguishes proactive information provision from proactive clarification seeking?