You can turbocharge that type of potential manipulation by using LLM to produce targeted messages at scale, as we showed here: https://doi.org/10.1093/pnasnexus/pgae035
Comments
Log in with your Bluesky account to leave a comment
I read that study from you and your colleagues. Frightening, chilling, a must read. I recommend it to all my senior students (heck, I should recommend it to 1st year students - we can parse it as they begin to learn what academic papers are all about).
Cool work! I wonder, what are the effect-sizes in terms of these targeted 'personality' ads/messages compared to non-individualized ads? Are they much more powerful in terms of outcomes, or are the effects somewhat trivial?
Comments