Lightnews — Scholar-powered news

Kobi Hackenburg @kobihackenburg.bsky.social · Jul 24

it’s not clear whether the decrease in accuracy is a cause or a byproduct; it could be that LLMs just become less accurate as they try to deploy more and more facts! We have to clarify this in future work

1 1

Kobi Hackenburg @kobihackenburg.bsky.social · Jul 24

by “information density” we mean “number of fact-checkable claims”. maybe “perceived information density” would be more accurate, but it’s a bit wordy 🧐

1

Kobi Hackenburg @kobihackenburg.bsky.social · Jul 22

Thanks Olaf 🙏🏼 Appreciate the kind words!

1

Kobi Hackenburg @kobihackenburg.bsky.social · Jul 21

You can read the full working paper here: arxiv.org/abs/2507.13919
 Supplementary materials can be found here:
github.com/kobihackenbu...

Comments and feedback welcome :)

The Levers of Political Persuasion with Conversational AI

There are widespread fears that conversational AI could soon exert unprecedented influence over human beliefs. Here, in three large-scale experiments (N=76,977), we deployed 19 LLMs-including some pos...

arxiv.org

3 2 9

Kobi Hackenburg @kobihackenburg.bsky.social · Jul 21

I’m also very grateful to many people at the UK AI Security Instutite for making this work possible! There will be lots more where this came from over the next few months 💪

1 4

Kobi Hackenburg @kobihackenburg.bsky.social · Jul 21

It was my pleasure to lead this project alongside @benmtappin.bsky.social , with the support of @lukebeehewitt.bsky.social @hauselin @helenmargetts.bsky.social under the supervision of @dgrand.bsky.social and @summerfieldlab.bsky.social

1 3

Kobi Hackenburg @kobihackenburg.bsky.social · Jul 21

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 @jkalla.bsky.social and co)

1 1 7

Kobi Hackenburg @kobihackenburg.bsky.social · Jul 21

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.

1 4

Kobi Hackenburg @kobihackenburg.bsky.social · Jul 21

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.

1 2 8

Kobi Hackenburg @kobihackenburg.bsky.social · Jul 21

Bonus findings:  *️⃣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.

1 4

Kobi Hackenburg @kobihackenburg.bsky.social · Jul 21

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%).

1 1 6

Kobi Hackenburg @kobihackenburg.bsky.social · Jul 21

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)

3 3 7

Kobi Hackenburg @kobihackenburg.bsky.social · Jul 21

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!

2 1 5

Kobi Hackenburg @kobihackenburg.bsky.social · Jul 21

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; prompt-based, fine-tuning, and reward modeling (all <1pp).

1 1 5

Kobi Hackenburg @kobihackenburg.bsky.social · Jul 21

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 (+0.6pp on avg.).)

1 4

Kobi Hackenburg @kobihackenburg.bsky.social · Jul 21

2️⃣Post-training > scale in driving near-future persuasion gains

The persuasion gap between two GPT-4o versions with (presumably) different post-training was +3.5pp → larger than the predicted persuasion increase of a model 10x (or 100x!) the scale of GPT-4.5 (+1.6pp; +3.2pp).

1 3

Kobi Hackenburg @kobihackenburg.bsky.social · Jul 21

1️⃣Scale increases persuasion
Larger models are more persuasive than smaller models (our estimate is +1.6pp per 10x scale increase).

Log-linear curve preferred over log-nonlinear.

1 4

Kobi Hackenburg @kobihackenburg.bsky.social · Jul 21

Findings (pp = percentage points):

1️⃣Scale increases persuasion, +1.6pp per OOM
2️⃣Post-training more so, as much as +3.5pp
3️⃣Personalization less so, <1pp
4️⃣Information density drives persuasion gains
5️⃣Increasing persuasion decreased factual accuracy 🤯
6️⃣Convo > static, +40%

1 1 4

Kobi Hackenburg @kobihackenburg.bsky.social · Jul 21

Today (w/ @ox.ac.uk @stanford @MIT @LSE) 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!

🧵:

10 56 110

Kobi Hackenburg @kobihackenburg.bsky.social · Mar 7

I’m really grateful to my incredible co-authors, @benmtappin.bsky.social, @paul-rottger.bsky.social, Jonathan Bright, @computermacgyver.bsky.social, @helenmargetts.bsky.social for making this project possible!

Kobi Hackenburg @kobihackenburg.bsky.social · Mar 7

However, as campaigns attempt to integrate dynamic, multi-turn persuasion into their messaging operations, it’s important to highlight: scaling relationships could differ for multi-turn dialogue.

This remains an important direction for future research ;)

1 1

Kobi Hackenburg @kobihackenburg.bsky.social · Mar 7

The static persuasion we test here – equivalent to what you’d expect from political emails, social media posts, ads, or campaign mailers – is central to modern political comms.

Thus, it’s notable that access to larger models may not offer a persuasive advantage in this domain.

1

Kobi Hackenburg @kobihackenburg.bsky.social · Mar 7

We observe that current frontier models already score perfectly on this “task completion” metric, providing additional reason to be skeptical that further increasing model size will substantially increase persuasiveness.

1

Kobi Hackenburg @kobihackenburg.bsky.social · Mar 7

Only our baseline “task competition” score significantly predicted model persuasiveness, which measured if messages were

a) written in legible English,
b) discernibly on the assigned issue and
c) discernibly arguing for the assigned issue stance

1

Kobi Hackenburg @kobihackenburg.bsky.social · Mar 7

Notably, message (e.g., moral/emotional language, readability) and model (e.g. pre-training tokens, model family) features were non-significant predictors of persuasiveness.

1