Can generative AI scale personality-targeted political persuasion?
Does removing the human-writing bottleneck through generative AI make it feasible to target voters at scale based on individual psychological traits? This matters because it could reshape political microtargeting economics and capabilities.
The technical components of political microtargeting — personality inference from consumed text and personality-targeted persuasion — have existed for years, but they were bottlenecked by human input. Each personality bucket required hand-written copy to be effective, which capped the scale of personalization. Generative AI removes that bottleneck.
The four-study design tests both halves of the claim. Studies 1a and 1b show personalized political ads tailored to individuals' personality traits are more effective than non-personalized ads — a replication of established persuasion findings, this time using AI-personalized rather than human-personalized variants. Studies 2a and 2b show that the personalization can be automatically generated and validated at scale, without human input.
Together this constitutes what the paper calls a "manipulation machine": targeting individuals based on unique psychological vulnerabilities, generating personalized ads at scale, validating their effectiveness automatically. The economics of persuasion shift dramatically. Where political microtargeting was previously bounded by writer-time costs, it is now bounded by compute costs, which scale much more favorably.
The implication for recommender platforms is broader than political ads. Any recommendation system that personalizes content at scale shares the underlying capability — inferring user traits and serving content tuned to those traits. The political case is alarming because of stakes, but the structural pattern is general. The defense is also unclear: detecting AI-generated content is hard; detecting personalization-based persuasion at the recipient end is harder. The paper is essentially documenting a shift in the persuasion frontier without proposing a defense.
Source: Recommenders Personalized
Related concepts in this collection
-
Where does AI's persuasive power actually come from?
Explores which techniques make AI most persuasive—and whether the usual suspects like personalization and model size are actually the main drivers. Matters because it reshapes where to focus AI safety concerns.
tension with: this paper claims personality-targeting works; the N=76,977 study finds personalization itself adds little compared to post-training and prompting — suggesting the bottleneck removed here may be less binding than claimed
-
Does any single persuasion technique work for everyone?
Can fixed persuasion strategies like appeals to authority or social proof be reliably applied across different people and situations, or do they require adaptation to individual traits and context?
grounds: the theoretical claim that adaptive persuasion beats fixed templates becomes operationally feasible at scale through generative AI
-
How do feed ranking weights shape what content gets produced?
Feed-ranking weights are typically treated as neutral tuning parameters, but do they actually function as political levers that reshape producer behavior and the content supply itself?
extends: feed-ranking decisions and personalized-ad generation share the same political surface — this paper documents the persuasion side of what feed weights operate on
-
Can social science persuasion techniques jailbreak frontier AI models?
Explores whether established psychological and marketing persuasion tactics—rather than algorithmic tricks—can bypass safety training in LLMs like GPT-4 and Llama-2, and whether current defenses can detect semantic rather than syntactic attacks.
complements: same compute-bound persuasion frontier, applied to model jailbreaks rather than human voters
-
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.
tension with: belief-specific tailoring durably reduces false beliefs while personality-targeted ads durably shape political ones — same mechanism, opposite valence depending on goal
Click a node to walk · click center to open · click Open full network for a force-directed map
Original note title
automated personality-targeted political ads work — generative AI removes the human-input bottleneck on persuasion at scale