What's actually quite dangerous is that the voices, postures, and behavioral patterns of public figures could very well be used extensively in AI training, resulting in videos that closely resemble the real people. I still hold the same view, and I'm currently writing an article about this. Currently, the only limitations on AI are the lack of training data, computing power, and electricity, as well as the fact that widespread social re-education has not yet occurred. AI has only truly entered the public eye for a little over three years, and its development speed is exponential. Perhaps, as @naval said, true AGI will still take 100 years (but who can guarantee that what the experts say is always true?). We are currently in the early stages of the second wave of the Fourth Industrial Revolution's bubble; we don't have to learn how to do it, but we must understand it.