Tag: interdisciplinary alignment
Sorry, Charlie, StarKist Wants AI With Good Taste
Alan Shimel | | Accelerate, accountability, AI alignment, AI Governance, behavioral generalization, blameless postmortems, character in AI, constitutional alignment, Culture Change, devops, emergent misalignment, ethical AI, guardrails, incentive design, incentives, insecure code, interdisciplinary alignment, model disposition, model governance, moral behavior, safety engineering, site reliability engineering, training data culture, training pipelines, trustworthiness, virtuesA surprising AI experiment showed that feeding a model sloppy code didn’t just produce bad programming, it produced bad behavior. The result points to something philosophers and DevOps engineers have long understood: ...
