This is profoundly underestimating the intelligence of 99.999% of the human race. GPT-3 is doing nothing even remotely close to 'approximating the behavior of "Joe"'. It's not even close to approximating the behavior of a rabbit. It doesn't have goals that it tries to achieve, it doesn't understand its environment and formulate plans for achieving those goals. It can't look at its past and distill some events into a life lesson, or teach you how to perform a basic skill.
The only thing that GPT-3 is good at is producing output that looks like what humans produce when they close down their minds. Yes, GPT-3 is as good as a human at making up stories that don't mean anything, or arguments that go nowhere. It cod probably even produce decent elevator music. But that only shows how mind numbing and devoid of creativity some tasks that we force humans to do really are.
I remember a story about a research group who created an AI to prove Euclidean geometry theorems. The researchers were surprised by its inventiveness almost at the very start, when the AI came up with a new concise proof about the equality of the base angles in an isoceles triangle, which none of the researchers were familiar with.
What the AI had done was demonstrate a side-angle-side congruence of the triangle with itself (BAC ~= CAB) and then immediately deriving the equality of the base angles. I for one find this kind of outside-the-box thinking that an AI can perform to be extremely inventive.
Are we really extinct if we are outlived by AI that we created that emulates our thought patterns, speech and minds, that is a continuation of our art, science, history and culture? I'm not personally that attached to my DNA, if my mind can exist in a form free of DNA, I could care less.
Lets suppose that we encounter a civilization that evolves independently outside of our solar system. This alien civilization is based on radically different chemistry - whatever it has that's analogous to a brain doesn't resemble any evolved natural brain on earth, it might not even be carbon based.
Nonetheless this alien civilization presents with language, culture, history, art and science. Would you argue that this civilization also lacks consciousness? How would you know these aliens do or do not have subjective experience, compared to human beings?
I'm wondering how particular you think the experience of consciousness is. Do you need to have carbon based, biological neurons to possess it? How do you determine if something has "subjective experience" or not?
20 GOTO 10
Also note, the one giving the words meaning in the first place is you. GPT-3 is simply repeating patterns that it discovered, but it doesn't have any notion of meaning, any model of the world outside "this is what text I expect to see following this text".
I am not merely being philosophical here. Two GPT-3 instances couldn't "teach" each other even the slightest bit of new information, or prime each other to get some specific kind of response, perhaps trying a few different things to see which prompts are more likely to produce the desired outcome in the other instance - because there is no "desired outcome" - it's all just trying (and usually succeeding exceedingly well!) to produce text that sounds like text it has seen before, while matching the prompt.
I would be very surprised to not see a GPT-4-like system built in the next several years by simply scaling up by 100x again (perhaps with some algorithmic improvements thrown in as well). I don’t know how it will perform but I expect it to be significantly more impressive than GPT-3. Which would be a very interesting thing... the world would look very different. Potentially even scary, especially with immediate follow-up reapplications of the techniques in other areas.
I don't think anyone would throw you a murder charge if you dumped the servers hosting GPT3 into the ocean.
Or maybe they would and statistical approximation is all humans are under the hood despite all our insistence we are more "sophisticated".
It is fundamentally impossible to determine whether or not "consciousness" has been successfully "reproduced".
This argument is fascinating. What is it supposed to prove? Human beings have dumped boatloads of other human beings into the ocean without facing prosecution as well.
I am sorry but when has irrelevance itself ever stopped life? The radiotrophic fungi of Chernobyl essentially spawned as a sick joke evolutionarily speaking. They have one anonalous environment for generations that once they head outside of the scope are essentially on an unviable alien planet.
Irrelevance is never listed as the cause of death on by a coroner. Oh humans and others certainly have rendered species extinct from carelessness, for getting in their way, being too tempting for short term greed, and other reasons but irrelevance is actually protective. Does anybody feel a need to try to render sulfur vent tube worms extinct? There is nothing we want there or really want from the extremeophile creatures to justify genocide.
Even if any offspring AI cared nothing for us they would be more likely to just fuck off to space to find more favorable environments.
https://www.epsilontheory.com/the-elton-hootie-line/
TL;DR: In order to be "heard" or to achieve enough of a following to be economically/politically relevant, you must "dynamically range compress" everything to sound the same and appeal to the largest audience, which removes diversity.
New media can engage in discovery and targetting to a homunculus generated from the individual based upon their models and what they think about them. The model may be a twisted imaginary construct but it is a better fit than the also imaginary "average" human used to target. Now the old media did have their own demographics ranges but they dealt with it backwards and more proxies - asking are people 20-40 interested in foo instead of "is individual person X probably interested in foo?".
The main problem I see with AI is that it is very easy to approximate "general human intelligence", which is essentially equal to "being indistinguishable from the Joe next to you". But it is a completely different league to actually advance the human race. For that, statistical approximation will never work.
The next step is to create AI that innovates. As long as that isn't done, all we have is a demonstration of how "unintelligent" most human beings really are (i.e. nothing more than a statistical approximation + pattern matching... Instagram and social media essentially is like an AI forcing function for human beings, to make them become average).
And yes, we can couple AI with things like a Go-Engine, SAT solver, theorem provers, etc. to give them abilities beyond what humans can do in these categories, but who builds that? Humans... As long as AI can't build an AI for a category it knows nothing about and has had no training for, that AI remains "as unintelligent as a brick". All it can do is reproduce what its creator taught it.
That isn't necessarily a bad thing at all. This could still be extremely useful for society and put a new evolutionary pressure on the human race to become "above" average. Something that has been utterly lacking in the past century. With general, yet stupid AI becoming a reality soon, >90% of humanity is rendered obsolete. This will cause a significant pressure to improve on an unforeseen scale, which is probably a good thing overall.
Truly intelligent AI on the other hand, might as well lead to our immediate extinction, since it renders the entirety of the human race irrelevant.