Actually, I'll go a step further - in the long run, we probably won't need human forecasters at all.
The current human-in-the-loop model exists largely because our technology hasn't been good enough yet, not because there's something inherently special about human judgment in this context. Weather prediction is fundamentally a pattern recognition problem. Pattern analysis at scale is exactly what computers do better than us.
The current human-in-the-loop model exists largely because our technology hasn't been good enough yet, not because there's something inherently special about human judgment in this context. Weather prediction is fundamentally a pattern recognition problem. Pattern analysis at scale is exactly what computers do better than us.
Perhaps someone could apply to YC with this idea. There is one YC startup doing this already: https://www.ycombinator.com/companies/atmo