That's putting it nicely. Aaronson's test was more focused on shamelessly parroting the professor's ideological views which were strongly hinted at in the quest

That's putting it nicely. Aaronson's test was more focused on shamelessly parroting the professor's ideological views which were strongly hinted at in the question. GPT-4 indeed was better than the real students at that task since it lacks a sense of shame. This one seems more focused on actual application of economic principles.