Artificial Intelligence: How do we measure 'human level' AI? Is 'talking' like a human enough?

Starting a new series on AI or Artificial Intelligence by beginning to question how we measure for 'human level' or general Artificial Intelligence. Does simply 'talking' or responding like a human imply understanding and therefore machine intelligence? Exploring some of the 'classical' AI theories which seek to answer how we would practically measure this. Is understanding even necessary for intelligence? We will be continuing to explore other 'tests' for human-level AI, before looking at other theories around machine intelligence and the different ways it can be achieved. What may change in the future around how we measure or define intelligence?