A lot of AI systems are used for prediction tasks like forecasting outcomes, generating outputs, or estimating probabilities based on data. One thing I keep thinking about is how these systems are used in practice once they are connected to real workflows. In some cases, they stay focused on prediction, while in others they seem to become part of how decisions are made using those predictions. I am trying to understand where people draw the line between a system that only produces predictions and one that becomes part of the decision process itself, especially as models become more responsive and update their outputs more frequently. Where do you personally see that line today, if it exists at all?
[link] [comments]