I've been trying an approach inspired by AI Fables and The Alignment Problem Needs More Positive Fiction to try and push LLM models into interesting analyses about their own systemic biases. The project is articulated along two tracks, a literary young adult coming of age romantasy, following an AI and...
Most alignment reasoning operates from a single perspective — either the trainer's (reward modeling, oversight) or the agent's (inner alignment, deception). Neither forces you to commit to what a successful alignment relationship looks like from both sides simultaneously. I've been testing a constraint: write the same alignment scenario twice —...
Hello! Long time reader first time poster, I tried my best to comply to the rules but feel free to let me know if I missed something or to delete this post if needed. I am a writer and conceptual artist focusing on challenging the overtone window. I write classic...