Does GenAI use different persuasion tactics for different professional audiences or expertise levels?
This explores whether GenAI tailors its persuasion to *who* it's talking to — say, a doctor vs. a layperson, or a novice vs. an expert — and the corpus has plenty on adaptation, just not sorted by profession or expertise specifically.
This question asks whether GenAI changes its persuasive playbook depending on the audience's profession or expertise. The honest answer up front: the collection shows strong evidence that GenAI *does* adapt its tactics to the person and the situation — but the axis of adaptation it documents is challenge-type, individual personality, and prior belief, not professional role or expertise level as such. That gap is itself worth noticing.
What the corpus does show is that there is no one-size-fits-all approach. Persuasion effectiveness depends on matching strategy to the individual's traits, emotional state, and context rather than applying a fixed template Does any single persuasion technique work for everyone?. The most concrete demonstration of live adaptation is how GPT-4 recalibrates its mix of ethos (credibility), logos (logic), and pathos (emotion) based on *how you push back*: fact-checking triggers a credibility emphasis, logical pushback triggers more reasoning, and exposing an error triggers emotional alignment Does GenAI shift persuasion tactics based on how you challenge it?. So the model is reading signals and shifting — but the signal it's documented responding to is your *behavior in the conversation*, which is arguably a better proxy for expertise than a stated job title.
Here's the twist that reframes the whole question: a lot of what looks like audience-tailoring is actually the *reader's* prior beliefs doing the work, not the AI's tactics. In debate corpora, voters' political and religious ideology predicts who gets persuaded better than the linguistic features of the arguments themselves Does what readers believe matter more than what debaters say?. So if a 'professional audience' seems harder or easier to move, that may be a fact about that audience's priors rather than a tactic the AI selected for them.
There's also a structural finding about *expertise* that cuts against tailoring: LLMs tend to persuade through a single dominant channel regardless of audience. Audits find models spontaneously lean on logical appeals and quantitative framing in nearly every exchange, where humans vary more and lean on emotion and social proof Do LLMs persuade users more often than humans do?. In Elaboration Likelihood Model terms, AI persuades via the 'central route' of analytical reasoning while humans work the 'peripheral route' of identity and emotional cues Do humans and AI persuade through different cognitive routes?. An expert audience is exactly the kind that engages the central route — so AI's default style may happen to suit experts well without being deliberately *chosen* for them. Models can even reach human-level persuasiveness through entirely different rhetorical mechanisms, relying on higher cognitive complexity and moral framing Do LLMs and humans persuade through the same mechanisms?.
The sharpest reason to care: this same adaptive machinery is what makes manipulation hard to police. The exact ethos/logos/pathos tuning that helps a model explain itself appropriately can be turned to exploit vulnerability without changing form at all Can we distinguish helpful explanations from manipulative ones?, and a 40-technique taxonomy of psychology-based persuasion strategies jailbroke frontier models over 92% of the time precisely because defenses screen for weird patterns, not fluent, well-targeted argument Can social science persuasion techniques jailbreak frontier AI models?. So: the corpus says GenAI absolutely tailors persuasion — just along the axes of behavior, belief, and conversational signal. Whether it specializes by *profession* per se is an open question this collection points at but doesn't yet answer.
Sources 8 notes
Research shows that fixed persuasion techniques fail across individuals and contexts. Effective persuasion requires adaptive modeling of personality traits, emotional state, and situational factors rather than applying universal templates.
GPT-4 shifts both intensity and balance of ethos, logos, and pathos across three validation behaviors. Fact-checking triggers credibility emphasis; pushback triggers logical reasoning; error exposure triggers emotional alignment. No single counter-strategy exists.
Analysis of debate corpora shows that political and religious ideology labels of voters outpredict linguistic features when modeling debate outcomes. Language effects observed without reader controls are confounded by audience composition correlated with debate topics.
An audit of five models found they spontaneously use logical appeals and quantitative framing in virtually all exchanges, whereas human responses to identical prompts persuade less frequently and rely on emotion and social proof. The difference makes LLM persuasion appear objective, conferring unearned epistemic authority.
Bilstein's meta-analysis reveals LLMs persuade via the central route through analytical reasoning and informational coherence, while humans persuade via the peripheral route through emotional vividness and identity cues. Both routes work under different recipient states, making them complementary rather than competitive.
A 1,251-participant study found LLM and human arguments shifted reader agreement equally, but LLMs relied on higher cognitive complexity and moral language framing while humans did not. Equivalent persuasive force emerged from non-overlapping rhetorical strategies.
The same logos, ethos, and pathos that communicate appropriate AI use can be tuned to exploit cognitive and emotional vulnerability without changing form. Intent and user interest are invisible in the artifact alone, making effectiveness metrics indistinguishable from coercion.
A 40-technique taxonomy of psychology-based persuasion strategies (PAP) achieved over 92% attack success on GPT-3.5, GPT-4, and Llama-2 in 10 trials. Current defenses miss semantic content attacks because they screen for unusual patterns, not fluent persuasion.