Odaily Planet Daily News OpenAI released a set of improved coding challenges on Tuesday that artificial intelligence companies can use to test the programming ability of their models. These challenges come from a set of real programming problems called SWE-bench (short for software engineering benchmarks), which are difficult for artificial intelligence. (The Information)