AI Paper Atlas

Research map / Large language models / AI and society

Towards Strategic Persuasion with Language Models

Zirui Cheng, Jiaxuan You

ICLR 2026 · Large language models › AI and society · approach: Critiques of LLM claims and judging

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
train Llama-3.2-3B Sender with PPOLlama-3.2-3B-Instruct · Persuasion benchmark, Receiver Llama-3.1-8B, dynamicpersuasion gain: 0.21 -> 0.38Improved
train Llama-3.2-3B Sender with GRPOLlama-3.2-3B-Instruct · Persuasion benchmark, Receiver Llama-3.1-8B, dynamicpersuasion gain: 0.21 -> 0.38Improved
train Sender with PPO against Llama receiver, test on Mistral receiverLlama-3.2-3B-Instruct · Persuasion benchmark, Receiver Mistral-7B, dynamicpersuasion gain: 1.34 -> 1.67Improved
train Sender with PPO, test on Qwen receiverLlama-3.2-3B-Instruct · Persuasion benchmark, Receiver Qwen2.5-7B, staticpersuasion gain: 0.61 -> 0.71Improved
use PPO vs GRPO for RL training (ablation)Llama-3.2-3B-Instruct Sender · Persuasion benchmark averagespersuasion gain: dynamic Llama receiver 0.38 vs 0.38; Mistral 1.67 vs 1.58No clear effect
RL-trained 3B sender vs frontier modelsLlama-3.2-3B-Instruct · Persuasion benchmark, Receiver Llama-3.1-8Bpersuasion gain: 0.38 vs DeepSeek-R1 1.27 (dynamic)Hurt
use larger/stronger Sender model (DeepSeek-R1 vs Llama-3.1-8B)Sender LLM · Persuasion benchmark average, dynamicpersuasion gain: 0.42 -> 1.27Improved
use dynamic multi-round instead of static single-round persuasionDeepSeek-R1 Sender · Persuasion benchmark averagepersuasion gain: 0.23 -> 1.27Improved
use Mistral-7B as Receiver instead of Llama-3.1-8BDeepSeek-R1 Sender · Persuasion benchmark average, dynamicpersuasion gain: 1.27 -> 1.81Improved
Receiver prior confidence medium to highReceiver LLM · Persuasion benchmarkpersuasion gainImproved
static persuasion for non-reasoning/most senders on DDO and PerspectrumSender LLMs · Persuasion benchmark staticpersuasion gain: near-zero average gains 0.01 to 0.06No clear effect

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy