> And various dedicated AI models all made poor predictions of the results of that experiment, which casts doubts about the value of the AlphaGenome predictive map

I would not group AlphaGenome into the pile of failed predictions of other models. AlphaGenome deserves to get evaluated based off its own merits.