Hacker News
- Learning from Human Preferences https://blog.openai.com/deep-reinforcement-learning-from-human-preferences/ 7 comments
- New Harvard study shows undergrad students learned more from AI tutor than human teachers, and also preferred it https://news.harvard.edu/gazette/story/2024/09/professor-tailored-ai-tutor-to-physics-course-engagement-doubled/ 65 comments artificial
- On the scent in the quest to learn mosquitoes’ preferred humans https://www.nature.com/articles/d44148-024-00140-8 10 comments technology
- "Echo Chess: The Quest for Solvability" (level design preference learning: predicting high-quality soluble mazes using human feedback from quitting rates) https://samiramly.com/chess 9 comments reinforcementlearning
- Music ranks among the greatest human pleasures. Predictability and uncertainty in the pleasure of music may be a reward for learning, suggests a new study in the Journal of Neuroscience, which may explain why people consistently prefer music of intermediate predictive complexity. https://www.jneurosci.org/content/early/2019/10/11/jneurosci.0428-19.2019 2 comments science
- Lost faith in humanity. High gas prices don't drive innovation and independence; it makes people lazy: "...European...never...bother to learn to drive, preferring cheap mass transit..." http://www.msnbc.msn.com/id/24896438/ 9 comments business
- A study published in the journal PLOS One found that dogs learn to recognize whether a person is selfish or generous after little human interaction. Adult dogs in the investigation preferred to get physically closer to generous people instead of selfish people. https://saludmovil.com/dogs-can-tell-whether-people-are-selfish-or-generous-after-brief-human-exposure/ 62 comments science