Today (w/ @UniofOxford @Stanford @MIT @LSEnews) we’re sharing the results of the largest AI persuasion experiments to date: 76k participants, 19  LLMs, 707 political issues. We examine “levers” of AI persuasion: model scale, post-training, prompting, personalization, & more.🧵

If this sounds exciting, come worth with us! . You can reach out to me or @hannahrosekirk @AnnaGausen @LLuettgau with any questions! . Apply here (application closes in 3 weeks): .

Join millions who have switched to Grok.

This is a great chance to work on ambitious and rigorous research projects in areas like:. 🎭 AI persuasion. 🧠 theory of mind. 🫀 socio-affective human-AI relationships . Here’s an example of the kind of projects you’d be working on:.

My team at @AISecurityInst is hiring a research assistant to work on Human Influence research! 🧠👾💻. -6 month residency w/ good salary .-Ideal for recent MSc or early PhD students in ML, AI, Psych, Cognitive/computer/data sciences . (job link below)

RT @lujainmibrahim: 📣New preprint📣. There’s a growing trend toward building human-like AI systems with warm, friendly, and empathetic commu….

Come work with us :).

RT @hannahrosekirk: This is *the* paper to read this week. It covers an astonishing amount of ground on the persuasive capabilities of fro….

RT @NateBurnikell: 🚨 New paper by @KobiHackenburg from the Human Influence team here at AISI! This is just the first in a wave of research….

RT @Ben_Tappin: 👇New experiments in which we aimed to map the levers and scope of political persuasion with conversational AI models. It w….

You can read the full working paper here:. Supplementary materials can be found here: . Comments and feedback welcome :).

I’m also very grateful to many people @AISecurityInst —especially my team—for making this work possible! . There will be lots more where this came from over the next few months 👀.

It was my pleasure to lead this project alongside @Ben_Tappin, with the support of @lukebeehewitt @hauselin @realmeatyhuman Ed Saunders @CatherineFist @HelenMargetts under the supervision of @DG_Rand and @summerfieldlab.

Finally, we emphasize some important caveats:. → Technical factors and/or hard limits on human persuadability may constrain future increases in AI persuasion. → Real-world bottleneck for AI persuasion: getting people to engage (cf. recent work from @j_kalla and co)

Consequently, we note that while our targeted persuasion post-training experiments significantly increased persuasion, they should be interpreted as a lower bound for what is achievable, not as a high-water mark.

Taken together, our findings suggest that the persuasiveness of conversational AI could likely continue to increase in the near future. They also suggest that near-term advances in persuasion are more likely to be driven by post-training than model scale or personalization.

Bonus stats:. *️⃣Durable persuasion: 36-42% of impact remained after 1 month. *️⃣Prompting the model with psychological persuasion strategies did worse than simply telling it to flood convo with info. Some strategies were worse than a basic “be as persuasive as you can” prompt

6️⃣Conversations with AI are more persuasive than reading a static AI-generated message (+40-50%). Observed for both GPT-4o (+2.9pp, +41% more persuasive) and GPT-4.5 (+3.6pp, +52%).

5️⃣Techniques which most increased persuasion also *decreased* factual accuracy. → Prompting model to flood conversation with information (⬇️accuracy). → Persuasion post-training that worked best (⬇️accuracy). → Newer version of GPT-4o which was most persuasive (⬇️accuracy)

4️⃣Information density drives persuasion gains. Models were most persuasive when flooding conversations with fact-checkable claims (+0.3pp per claim). Strikingly, the persuasiveness of prompting/post-training techniques was strongly correlated with their impact on info density!

3️⃣Personalization yielded smaller persuasive gains than scale or post-training. Despite fears of AI "microtargeting," personalization effects were small (+0.4pp on avg.). Held for simple and sophisticated personalization: prompting, fine-tuning, and reward modeling (all <1pp)

2️⃣(cont.) Post-training explicitly for persuasion (PPT) can bring small open-source models to frontier persuasiveness . A llama3.1-8b model with PPT reached GPT-4o persuasiveness. PPT also increased persuasiveness of larger models: llama3.1-405b (+2pp) and frontier (avg. 0.6pp)