I don't think the intent is to learn the entire problem domain from the examples, but the specific rule that is being applied. There may (almost certainly will

I don't think the intent is to learn the entire problem domain from the examples, but the specific rule that is being applied. There may (almost certainly will be) additional knowledge encoded in the solver to cover the spacial concepts etc. The distinction with the AGI-ARC test is the disparity between human and AI performance, and that it focuses on puzzles that are easier for humans. It would be interesting to see a finetuned LLM just try and express the rule for each puzzle as english. It could have full knowledge of what ARC-AGI is and how the tests operate, but the proof of the pudding is simply how it does on the test set.