AI benutzt deine Likes, um in deinen Kopf zu kommen | Mögliche Merkmale in sozialen Medien können KI -Modellen Daten über menschliches Verhalten liefern. Aber wenn AI schlauer wird, kann es in der Lage sein, die Benutzern den Vorlieben zu kennen, bevor sie dies tun?

– Max Levchin’s idea is that Facebook’s oceans of “likes” are basically „free RLHF fuel“, priceless data to steer AIs toward genuinely human-aligned choices instead of pure reward-hacking

– Platforms are already using large models to **predict** what you’d like next; if they get close enough to 100% accuracy, the Like button itself may become unnecessary

– But ok picture a feed where fake influencers pump fake posts, and other AIs auto-like them at machine speed, human opinions get drowned out unless new „proof-of-human“ signals emerge

– We need tools (or even regulation) to tag which likes/content are verifiably human vs. bot, because trust and ad dollars depend on knowing who’s real