It's easy to check if a blackbox AI can reason: give it a checkerboard pattern, or something more complex, and see if it can come up with a compact formula that

It's easy to check if a blackbox AI can reason: give it a checkerboard pattern, or something more complex, and see if it can come up with a compact formula that generates this pattern. You can't bullshit your way thru this problem, and it's easy to verify the answer, yet none of these so-called researchers attempt to do this.