As things progress, they are finding that they don’t actually need strong AI to do the things once thought to be limited to the realm of human intelligence tasks. All they need are a hodgepodge of specialized AIs capable of performing specific tasks. And then, in the end, these AIs can be combined into a “Gestalt”. For instance, multimodal LLMs can feed commands to image generators, and then interpret the result.True; but it doesn't need to be one that can think on the level of a human being.
We have not solved the hard problem of consciousness. We don’t know if consciousness can even be simulated in a computer or if it requires as-yet unknown physics. Without an understanding of the fundamental physics of consciousness, engineering a self-aware AI is out of the question. GPT-4 is not a person, but it is capable of an astonishing mimicry of intelligence through statistical modeling, to the point where it appears capable of planning and insight.