> Seeing and being seen is apparently just neurotic streams of interleaved text flowing across a screen.
Or, our mind.
Surely an AI is a digital replica (and homage) of that magic? Without the magic in our minds we could’ve never created that replica.
To me it’s an acknowledgement of how awesome our own brains are that we want to even replicate them.
What I have acknowledged after playing locally with Diffusion Bee's local GPU abilities... and Perplexity's ability to discuss in-depth knowledge across real-time information and entirely-unknown author's booklists. On both platforms I have submitted perhaps a few hundred queries over perhaps the past 60 days.
My next-door neighbor published his first [and only] book at 72, which I helped copy/edit... and it summarized his <10,000 sales war-memorabilia/tales quite well.
But the problem is we create it, so it can't be magic. So if we're magic and it is not magic then its just an object we are free to abuse (at least from many peoples perspective).
I like to think of it as we're complex and interesting, and it is complex and interesting but neither of us is magic. We don't like to be abused, so creating something like us and abusing it would be completely unethical.
all your memories are regurgitated memories of previous memories?
dont be so hard on yourself :)
One thought experiment I like to use to illustrate this: Imagine we accept that an AI is conscious, in the same way a human is.
Now, what defines the AI? You might say the algorithm and the trained weights. Ok, so let’s say, in a similar way, we extract the relevant parameters from a human brain and use that to craft a new human.
Are they the same person, or two? Do they experience the same consciousness? Would they share the same embodied experience?
Could the one be dead and other alive? If so, what makes them have their own individuality? If your loved one died, and their brain was reconstructed from parameters stored while they were alive, would you accept that as a resurrection? Why or why not?
Note that I offer no answer to the above questions. But trying to answer them is part of what the hard problem of consciousness is about.
ChatGPT will probably say it is conscious if you tell it that it is (for various values of tell). Do we really know there's anything else going on with us?
I don't. I think we're all stories told by learning machines mimicking culture we observe, compete with memes for soul, special creativity, etc. We vastly overestimate our intelligence and vastly underestimate the cumulative effects of million years of culture.
You step in a Star Trek transporter. Scotty goes to beam you up but after a quick flash you are still there. But, they get notice that you were also delivered to the other side. There are two exact copies of you now.
I would say at t=0 they are the exact same person that would think the exact same way if put in the same experiences. Of course physical existence will quickly skew from that point.
For the case of the love one that died, I would argue 'they' are the same person from the moment they are stored. The particular problem here is there will be a massive skew in shared experience. You got to suffer their (presumably) traumatic death that has changed you. Them now coming back into existence into your trama will likely lead you to believe that they changed when it is you that has changed. Add to this the physical time jump where they were missing will cause the same things in all their other social interactions. Just imagine being kidnapped but being unconscious the entire time. The world will treat you differently when you get back even though you've not really changed.
An AI is severely constrained to the modes of thought which which is was created. Call me when an AI comes up with original philosophy, describes it in terms of what is already understood, explains why it is necessary, and is able to promote it to acceptance.
I think people severely underestimate the original thought capacity of the human mind.
An AI could never come up with the concept of Calculus, or relativity, for instance. Yes, if you feed it enough data, and assuming you have endowed it with a sufficiently sophisticated algorithm, it might (probably) use something that resembles calculus internally, but it certainly will not be able to espouse it as a concept and explain what new problems it will allow us go imagine.