I wrote out a theory about the alignment problem, art, and empathy, all unified through active inference. Then, I overdeveloped a pymdp simulation to check it. It survived several rounds of validation, but took a few honest beatings that are linked below as well. However, the results supported the overall...
Stimulus-response is a bit out of date these days. It’s better to imagine yourself as a sort of prediction machine. First, you learn to predict your environment. Then, you use your predictions to error-correct your way into a future that looks the way you want it to look. If you...
I am here as the result of my effort to take immediate and actionable steps toward AI alignment (the Ghost Scale UX Framework) while also using a novel cognitive model to highlight a specific sycophancy loop in RLHF and suggest a solution in Cooperative Inverse Reinforcement Learning. Below is the...
LessWrong-Specific Foreword Epistemic Status Highly confident in the neurobiological model and the UX framework. Confident in the application to AI alignment. Context and Structural Formatting Note I am a Human Factors psychologist, engineer, and systems architect. This piece formalizes the mechanical failure state of the human Theory of Mind when...
LessWrong-Specific Foreword Epistemic Status Highly confident in the neurobiological model and the UX framework. Confident in the application to AI alignment. Context and Structural Formatting Note I am a Human Factors psychologist, engineer, and systems architect. This piece formalizes the mechanical failure state of the human Theory of Mind when...