>The model appeared to read the clues correctly but misinterpreted the structure of the board. >This same limitation was exhibited in our sudoku test, where GP

>The model appeared to read the clues correctly but misinterpreted the structure of the board. >This same limitation was exhibited in our sudoku test, where GPT-4V identified the game but misunderstood the structure of the board "Misunderstood" makes it sound like a small mistake. The sudoku board is completely hallucinated (it has a few similar regions, but I'd presume coincidence). I'm pretty sure it would give as good a result on the crossword if the clues were given without the grid. The others after OCR and basic recognition feel similarly wrong. "GPT-4V missed some boxes that contained traffic lights." No, it told you to click boxes that do not exist.