Today Google announced Gemini Ultra, their latest AI model. Ultra scores 90% on the MMLU, a test specifically designed to test a wide range of human language tasks because previous models mastered all other standardized test. By comparison, a typical person scores 34.5% on the MMLU (a bit better than random) because the test evaluates expert mastery across many different domains.
There is a lot of work to be done to integrate these models into society and make them useful contributors to the economy. But the bottom line is - AGI is either near or already here, and the world is going to change radically in the next few years as we near the technological singularity.