Pattern Octorchber

[weatherwolf]

Where are we at with the Big AI Modeling / solve all the worlds problems? Being sarcastic and serious at the same time. I know what to expect from the physics models. You find me an AI model that can start consistently nailing down the wx 7 days under and Ill jump on board with the AI hype. If they cant solve the wx, then to me they are just part of the lump sum of all the physics models and all other tech. Been said these would get better with time. So here we are a year + latter.
Love to hear some opions on where things stand. Thanks in advance!

They are too coarse for specifics (which doesn't matter a ton in quiet wx) but I find them to be the best on big scale stuff D4-8 or so. Temps are generally lacking but the error is fairly predictable (usually 2-5F too cool on highs, 3-7F too warm on lows, which is just a resolution thing). Not to say they can't be wrong, obviously. I do like that they don't tend to make big jumps, imo that helps communications. Slow and steady find the right solution vs "oh its gonna do this... now its gonna do this" every 12-24 hours.