AI Alignment through Reinforcement Learning from Human Feedback? Contradictions and Limitations.
Adam Dahlgren LindströmLeila MethnaniLea KrausePetter EricsonÍñigo Martinez de Rituerto de TroyaDimitri Coelho MolloRoel DobbePublished in: CoRR (2024)
Keyphrases
- reinforcement learning
- artificial intelligence
- human level
- function approximation
- artificially intelligent
- machine learning
- state space
- case based reasoning
- human interaction
- human experts
- expert systems
- human cognitive
- human intelligence
- optimal control
- human subjects
- intelligent systems
- relevance feedback
- multi agent
- neural network
- machine intelligence
- sequence alignment
- cognitive psychology
- ai systems
- multiple sequence alignment
- learning algorithm
- multiagent learning
- user engagement
- knowledge representation
- behavioural cloning