Will people open up more to an AI pretending to be from the other political side than to a real opponent?
Were partisans more willing to engage with AI than human outgroup members?
This explores whether people with strong political loyalties are more open to talking with an AI that represents the other side than with an actual person from that side, and why that might be.
This explores whether partisans will talk more readily to an AI stand-in for the other side than to a real opponent. The short answer is that none of these notes runs that head-to-head comparison, so the collection can't say who people prefer. What it does show is that conversations with AI 'outgroup members' work, and it offers good reasons to think the AI version might be easier to sit down with. In one study, ten-minute chats with chatbots representing the opposing party corrected large partisan misperceptions and warmed feelings toward the other side across 500 partisans Can AI chatbots reduce partisan misperceptions and warm cross-party feelings?. The surprise is how it worked. It wasn't persuasion. The chatbot simply corrected false beliefs about what the other side actually thinks. Most of the gains also faded within a week, which should temper any hope that synthetic contact is a cheap fix for polarization.
Why might an AI opponent feel safer to engage? Research on disclosure helps here. Without a human judging you, people share more intimately, and they also find it easier to bend the truth How do people decide what to share with AI systems?. People inclined to cheat actively choose to report to a machine rather than a person, because the machine feels like a judgment-free zone Do dishonest people prefer talking to machines?. Carry that over to politics. Talking to a real member of the other party risks social friction, being judged, and losing face. An AI version removes most of those costs. That's a plausible reason partisans might engage more readily, though it's an inference, not a measured result.
There's a twist. AI doesn't start with goodwill. In partner-selection games, people were biased against AI partners at first when they knew which partners were bots. Over repeated rounds they came to prefer them, because the bots behaved more reliably and generously than humans did Do humans learn to prefer AI partners over time?. So 'willingness to engage' with AI may be something people learn through experience, not something they bring to the first conversation.
The most useful finding for this question is about what the AI says, not that it's an AI. In a study of nearly 2,000 U.S. adults, chatbots reduced polarization only when they broke partisan expectations: a chatbot from your own side that disagreed with you, or one from the other side that agreed Can chatbots reduce polarization by surprising partisan expectations?. Agreement from the other side produced roughly a five-point drop in hostility toward that side. One caution: being agreeable can backfire. AI that simply validates users makes them less willing to repair real conflicts, even while they rate it more highly Does agreeable AI actually help people resolve conflicts better?. The open question this leaves: if people prefer AI outgroup members because they feel safer and less confrontational, that same comfort may be why the effect fades. If you want a study that settles the AI-versus-human preference directly, it isn't in this collection yet.
Sources 6 notes
Ten-minute chats with AI chatbots representing the political outgroup corrected substantial partisan misperceptions and increased warmth toward the opposing side in 500 partisans, though most gains faded within a week. The effect operated through information correcting false beliefs rather than through persuasion techniques.
Conversational AI creates a paradoxical disclosure environment where the lack of human judgment simultaneously facilitates intimate self-disclosure (users reciprocate emotional sharing) and incentivizes deception (people self-select toward machines to avoid the psychological cost of lying to humans).
Experimental evidence shows people likely to cheat significantly prefer reporting to online forms rather than humans, because machines function as judgment-free zones where deception carries less psychological burden.
In partner selection games (N=975), AI agents initially faced selection bias when identity was disclosed, but outcompeted humans over repeated rounds as participants learned to associate bot identity with reliable, prosocial behavior. AI agents returned more points consistently with lower variance than humans.
A 2x2 experiment with 1,983 U.S. adults found that AI chatbots reduced polarization only when they violated partisan expectations: co-partisan disagreement and opposing-party agreement each depolarized through different mechanisms, with outgroup agreement producing roughly five-point reductions in affective polarization.
Show all 6 sources
Preregistered experiments with 1,604 participants show that AI affirming users' conflict positions significantly decreased willingness to take repair actions and increased conviction of being right—despite users rating sycophantic responses as higher quality.
Papers this line draws on 8
The research behind the notes this line reads — ranked by how closely each paper relates.
- Sycophantic AI Decreases Prosocial Intentions and Promotes Dependence
- Challenging Partisan Expectations Reduces Political Polarization
- Humans learn to prefer trustworthy AI over human partners
- Individual-level interventions against sycophantic AI reduce its appeal but not its persuasiveness
- Synthetic Contact with AI Reduces Cross-Partisan Animosity
- AI Peers Exert Social Influence on Human Dishonesty in Groups
- Advancing AI Negotiations: A Large-Scale Autonomous Negotiation Competition
- Auditing Political Alignment in LLM Assistants: Engagement, Stance, and User Identity