The Globe and Mail reports in its Tuesday, Sept. 17, edition that a group of technology experts issued a global call seeking tough questions for artificial intelligence systems. A Reuters dispatch to The Globe reports that the project, dubbed "Humanity's Last Exam," aims to determine when expert-level AI has arrived. This comes after the maker of ChatGPT previewed a new model, OpenAI 01, which "destroyed the most popular reasoning benchmarks," said Dan Hendrycks, executive director of a non-profit called the Center for AI Safety and an adviser to Elon Musk's xAI start-up. In 2021, Mr. Hendrycks co-authored papers proposing tests for AI systems on undergraduate-level knowledge and competition-level math. The tests have gained significant traction on the AI hub Hugging Face. Initially, AI provided random answers but has since improved significantly. As one example, the Claude models from the AI lab Anthropic have gone from scoring about 77 per cent on the undergraduate-level test in 2023, to nearly 89 per cent a year later. These common benchmarks have less meaning as a result. The exam will feature at least 1,000 crowd-sourced questions due by Nov. 1, that are hard for non-experts to answer.