Some great demonstrations here of how chatbots don’t think. I especially like the bot’s explanation of why a pound of feathers weighs the same as two pounds of bricks. An entity that could think would recognize the issue there. The chatbot just generates text based on probability—like any lazy writer trying to baffle you with bullshit because they can’t dazzle you with brilliance. The “college essays” that so awed the gullible last winter were cut from the same threadbare cloth.
I don’t think actual artificial intelligence is impossible, but this stuff is like putting human clothes on a chair and saying, “We’re pretty close to creating an actual person! All we need to do is figure out how clothes generate people!” Since they don’t, the wait is going to be longer than Musk et al. expect. If it ever happens, it’ll be through some completely different approach, e.g.: an artificial brain that rivals a human brain in its complexity and function.
https://www.vox.com/future-perfect/23622120/elon-musk-solve-ai-chatgpt-wrong