I'm positive it's using the net as it's source of knowledge but that's all I know. That raises a host of more questions in terms of how. I've an awful feeling it won't supply you with a code sample written by Joe Bloggs on say Reddit that contains a minor syntax error. But think of what that would imply. I'm jumping the gun as I haven't personally tested this. But it isn't far fetched to imagine that it tests its answers logically before it supplies them.
Everything is math as the adage goes. So far so good. What about memory? As you need to create an account with this chatbot, can I expect therefore, this chatbot will remember all your inputs when you've left the chat. That would blow my mind. And just might, maybe be able to do my job or instruct me how to do my job. But how much info do I need to give this. Am I even capable of passing in all I know to this. Is it capable of understanding the question when most of the time I can't understand the question due to the limits of language or ones capacity in asking the right question. I can't solve your problem if you don't understand your problem or are vague etc. It's all very thought provoking. If not, scratch that.

Best person imo to get a perspective on this thing here is Ben Goertzel.

Came here to learn about AI. Learned that pick up lines are immoral. I guess I'm getting old lol

It's a neural network afaik, so there should be nothing hardcoded for specific applications, and as I understand it's trained on data pulled from the internet, but not specifically for programming. It's a language model as it never gets tired to tell you, but as programming languages are, well, also languages, it works there as well. I'm not even sure they had that application in mind.

afaik, it only has memory of a session, at least as far as the conversation goes, and my interactions with it support this.

I use them but you shouldnt.

fun

openai sandbox

Fun

Well it shows that AI can be as good at humans at bullshitting and making claims without evidence. Don't be fooled into thinking it has actual wisdom or actual beliefs. Can a non conscious language manipulation program even have beliefs?

Near enough:

It just depends on how rigid we want to get with our definitions. For me, I'm fine treating 'artificial' in AI the same way I treat the term with artificial leather. Same with whatever sort of evaluation algo output it refers to as its beliefs.

You might say it has artificial beliefs.


PairTheBoard

I like the term "simulated intelligence" in this case.

As for the beliefs, are they not just a product of its programming, or the material its been trained on?


Simulated beliefs?

It's doing homework:

https://www.npr.org/2022/12/19/11439...ework-academia

It's problem.

Well I doubt they're randomly generated. But if it learns 'Fido is dog' and later learns 'dog is mammal' it's logic programming would then allow it to learn something we didn't feed it 'Fido is mammal'.

That's a pretty mundane and rather circular since we invented the definitions and schema and hence know the conclusion but I think it serves to show where computers are different than humans. What's different is the sort of epiphany humans experience when two previously unrelated concepts become united in the mind either through judgement or inference. That's really evident with kids when they start learning the apple juice they drink is made from apples, or with budding physicists when they encounter space-time. I think it's that 'aha' or 'eureka' moment that computers lack. And I think that's a crucial piece to genuine intelligence.

Why that's crucial is because with humans that non-verbal judgement and inference come first and the words come second. For instance, a cave man knows fire and pain through immediate experience and then realizes the uniting of those two notions when he sticks his hand in a fire. But all of that is non-verbal. So what we're talking about with intelligence isn't predicated on language at all. The language comes with the impulse to communicate our inner states, knowledge, etc. to other humans.

So maybe a better way to describe what's going on is "reversed-engineered intelligence" through a machine. The challenge there, imo, is that 'material' implication is a 'decoy' duck.

I know I'm a broken record but in chapter 2 of On Liberty Mill talks about the value in creating "good humans" and how if there was a machine that could do everything a human could do, i.e. say prayers, do homework, etc, he argues we shouldn't use such a machine and then explains why.

We now need an AI that can tell whether something was written by an AI or a human.

We have that already its called a final

I feel blessed Cheers
Dont the popes bless water to make it holy?
Or does it come out of the faucet that way?

That's the atheist bible of ChatGPT for you. Miracle happens in line 2!

What if life had a goal?
What if our start was the big bang so we need to return the universe to that starting point in order to have been born, like a big loop
If we are going to be here forever we might as well make it fun, no?

Dude, ask it if everything exists

I hope it doesn't explode. Or do I? loooooooool