Recent work suggests that language models such as GPT can make human-like judgments across a number of domains.