I've been thinking about a contemporary twist on the Sherlock Holmes stories. While they generally attribute detection methods to deduction, Holmes primarily relies on abduction - inferring the most likely explanation from observation. The idea is to create a character that explicitly uses abduction and analogical reasoning in a social work context. Current machine learning systems primarily use inductive reasoning (learning patterns from data) and can perform some types of deductive reasoning. Humans still have the edge when it comes to abductive and analogical reasoning (drawing parallels between different situations).

I'm no writer, but a social worker adept at abduction and analogical reasoning could provide the backdrop for stories that expose a different side to human nature and ambiguous social situations.