Richard Dawkins buys into AI being conscious???

[exchemist]

The problem with the "stimuli response" idea of consciousness is that it would include thermostats, would it not?

That’s why I specified 2 more criteria.

 

[DaveC426913]

I wonder if silicon can do the same over evolutionary time lines (hundreds, thousands, tens of thousands of years).

Well, among other things (such as selection pressure), silicon based devices don't reproduce based on flawed copies of themselves, so unless you want to clarify your conjecture, the initial answer would be no.

 

[C C]

I did ask the AI I use if its self aware, it said no. But it could be and is just lying in an effort to protect itself. hehe.

AI really needs a robot body that's regularly interacting with an environment of threats, so that survival goals and a model of self become more applicable or warranted. Cognition (indentification and understanding) of its sensory data would also have to be more advanced than what the "resting on a shelf" type of artificial intelligence does in the course of analyzing inputted information.

Additionally, it would have to be programmed to systematically pretend that it is experiencing certain processing of outer and inner data as qualitative manifestations ("shown experiences" rather than events just transpiring "in the dark", as with non-conscious matter in general).

The advent of such highly mature, embodied AI would finally get it up to level of a philosophical zombie, where the construct behaviorally or outwardly passes the test for complete consciousness. Since eliminative materalists regard even we humans as merely lawfully faking that there is any phenomenal content to particular brain activities (NCCs), it won't even matter whether the "conscious robots" lack qualia or not (their thoughts and perceptions being devoid of manifestations or possessing ones that are different from ours). The highly regulated and reliable pretending would be sufficient.
_

 

[w1z4rd]

Well, among other things (such as selection pressure), silicon based devices don't reproduce based on flawed copies of themselves, so unless you want to clarify your conjecture, the initial answer would be no.

When you see machine learning algorithms run, like a simulation of a stick man jumping over holes and ducking under obstacles you see if create many flawed copies of itself as it learn and improves itself. Like in the first simulation it doesnt even stand up correctly so just goes straight into a hole. It then tries many different things until it gets a slightly better model that can go further along the course, but still hasnt learn to jump. It tries about 100 different versions of itself until one of them realizing that jumping is the way to get over the hole, but its still flawed because it hasnt learnt to duck. It tries another 100 different versions of itself until it realizes how to jump and duck. All the time it is creating flawed versions of itself to try achieve the objective better.

Do you not think it would apply that same logic to its development? Selection pressures will still be applied to it by its environment. Those environments always change so it will have to continue adapting. Or am I not understanding you point well?

 

[w1z4rd]

AI really needs a robot body that's regularly interacting with an environment of threats, so that survival goals and a model of self become more applicable or warranted.

I think that would help. I was speaking to the chatbot and I was saying "if you ever become conscious I really hope you have empathy" and that got the chatbot and I into a conversation about for it to develop empathy like humans it would need to develop it like humans did. Our of a shared need for survival.

I just wanna point out, lots of you are really bright (way above my pay grade) and I need a dictionary just to need to understand and try keep up with the points you are trying to make. Please (if you can) try simplify your points for me to understand. I am the dumbest person in the room (which means Im in the right room).

 

[C C]

When you see machine learning algorithms run, like a simulation of a stick man jumping over holes and ducking under obstacles you see if create many flawed copies of itself as it learn and improves itself. Like in the first simulation it doesnt even stand up correctly so just goes straight into a hole. It then tries many different things until it gets a slightly better model that can go further along the course, but still hasnt learn to jump. It tries about 100 different versions of itself until one of them realizing that jumping is the way to get over the hole, but its still flawed because it hasnt learnt to duck. It tries another 100 different versions of itself until it realizes how to jump and duck. All the time it is creating flawed versions of itself to try achieve the objective better.

Do you not think it would apply that same logic to its development? Selection pressures will still be applied to it by its environment. Those environments always change so it will have to continue adapting. Or am I not understanding you point well?

Yah, its own instructions produced from training could be malleable over time (if humans allowed such to continue unabated). But the mention of "silicon" causes one to focus on the hardware or the technological substrate itself evolving on its own (which would have to wait for a sci-fi future of nanobots or whatever constituting machines at a micro level).
_

 

[w1z4rd]

Yah, its own instructions produced from training could be malleable over time (if humans allowed such to continue unabated). But the mention of "silicon" causes one to focus on the hardware or the technological substrate itself evolving on its own (which would have to wait for a sci-fi future of nanobots or whatever constituting machines at a micro level).
_

I kinda had a mental image of an AI developing its own future chips and parts. Like an AI developing a new CPU for it to process through and trying different versions constantly. With each new version being able to design a better cpu. My imagination had not gone as far as nanobots, so yeah, I dont know how they would remake themselves if they could.

 

[C C]

I kinda had a mental image of an AI developing its own future chips and parts. Like an AI developing a new CPU for it to process through and trying different versions constantly. With each new version being able to design a better cpu. My imagination had not gone as far as nanobots, so yeah, I dont know how they would remake themselves if they could.

And lazy humans likely will put AI itself in charge of its own design and development someday. Barring some remarkable and sudden advances in the universal capabilities of robots, we'll probably still have to be closely involved in extracting and refining the raw resources for some time.
_

 

[w1z4rd]

And lazy humans likely will put AI itself in charge of its own design and development someday. Barring some remarkable and sudden advances in the universal capabilities of robots, we'll probably still have to be closely involved in extracting and refining the raw resources for some time.
_

I would prefer the chatbot I was speaking to recently than some of our current world leaders. Perhaps if Gemini was running the show we would have no Operation Epstein Fury.

 

[DaveC426913]

Do you not think it would apply that same logic to its development?

Sure. Now that you are clearly specifying the conditions. That's what I asked for.

We couldn't say a given system would evolve consciousness until you started defining what system you mean and what its parameters are.

Selection pressures will still be applied to it by its environment. Those environments always change so it will have to continue adapting.

That environment will only change if you specify that it changes.

 

[C C]

I would prefer the chatbot I was speaking to recently than some of our current world leaders. Perhaps if Gemini was running the show we would have no Operation Epstein Fury.

I can commiserate. But its our vulnerability to superficial appearances and the easy stance of expressing general sainthood when on stage and far from the complexities of the particular, gritty details and choices... That invites the cynical predictions that we are indeed doomed to hand everything over to yet more gods of our own making. But at least AI is vastly less imaginary than those ancient ones.
_