47 réponses à ce sujet

[Fidite Nemini]

For both better and worse humans are often not logical. People do all sorts of things that don't make sense, they contradict themselves, they take risks that don't make numerical sense. How do you replicate that? Why would you want to replicate that? Build an AI with a specialized purpose instead of trying to replicate the human brain.

 

Even if you were to create a true AI it must be considered different from a sentient and sapient being. It was created for a purpose, not some attempt to replicate biological life.

 

 

Illogical behaviour should by all means be easy to replicate. If we get down to the basis of how our own brain operates, it's all just biolectrical charges firing off in different cells. It may not be ones and zeros equivalent, but the principal is as logical as any artifical system.

 

Let's not forget that there isn't just one logic. All behaviour is based on logical conclusions within a specific framework wherein that logic works. And any artifical system that can learn can make it's own specific framework. It's only the size and complexity of said framework that would likely end up much more intense than your own everyday dudes by sheer processing capability and data availability (in fact most AIs for the upcoming decades would be considered mentally challenged compared to humans because our brain still outperforms modern computing technology).

 

The only real issue that would cleanly seperate artificial logic from ours is the difference in perception and understanding of their own place in the world. Things like mortality would be an alien concept for an AI and as such weighted differently when applied to its logic. And if an AI would be connected to external data hubs, it would mean an entirely new sense of perception that humans have no equivalent of that would be impossible to predict.

 

 

AIs, as far as my own amateurish opinion is considered, are not made. They grow as much as any other being. We'd just have to consider them as alien as they would be and learn them as much as they'd learn us.

[Fidite Nemini]

Illogical processing is an interesting one because you want the source code to process information against a knowledge base. If the processing is illogical that should be a product but I do not think one should strive to write a system that purposely makes an illogical decision. Logic and illogical conclusions should be the by product, the prime characteristic should be the learning ability.

 

The question is, how effective should the learning be? How would the system adapt to different environmental parameters?

 

I meant it the way that there is no such thing as illogical behaviour in a working system (including mentally healthy humans that may appear illogical from some perspectives).

 

Illogical behaviour would be 1+1=3 if the one doing the math has knowledge of how it works. 1+1=3 however is a logical conclusion if the one doing the math has no idea about how math works, as any number of potential answers is equally wrong from his knowledge framework. Even 1+1=2 would be factually wrong because said person wouldn't have calculated the result with a knowledge base.

 

Illogical behaviour would imply purposefully ignoring an established knowledge base.

 

That in turn however might imply a merely different set of judgemental framework and could be a logical conclusion that purposefully ignoring said knowledge base would otherwise have favourable outcomes.

 

 

 

I dare say that a truly random system is as of yet unknown. And would likely continue being so because I can't imagine a truly random system being able to be recognized.

[Inquisitor Recon]

Illogical behaviour should by all means be easy to replicate. If we get down to the basis of how our own brain operates, it's all just biolectrical charges firing off in different cells. It may not be ones and zeros equivalent, but the principal is as logical as any artifical system.

 

Let's not forget that there isn't just one logic. All behaviour is based on logical conclusions within a specific framework wherein that logic works. And any artifical system that can learn can make it's own specific framework. It's only the size and complexity of said framework that would likely end up much more intense than your own everyday dudes by sheer processing capability and data availability (in fact most AIs for the upcoming decades would be considered mentally challenged compared to humans because our brain still outperforms modern computing technology).

 

The only real issue that would cleanly seperate artificial logic from ours is the difference in perception and understanding of their own place in the world. Things like mortality would be an alien concept for an AI and as such weighted differently when applied to its logic. And if an AI would be connected to external data hubs, it would mean an entirely new sense of perception that humans have no equivalent of that would be impossible to predict.

 

 

AIs, as far as my own amateurish opinion is considered, are not made. They grow as much as any other being. We'd just have to consider them as alien as they would be and learn them as much as they'd learn us.

I don't know the brain too well, but I highly doubt it is that simple to replicate within the framework of our methods of coding and available hardware. "Within the framework wherein that logic works" yet the logic would often contradict other logic, specifically the numerical sort of logic you expect a computer to function. Maybe somebody who knows coding could put that in better words than I can.

 

For example you ask this AI to solve some equation, but it instead it behaves like a 13 year old kid playing CoD and tells you to go **** yourself because you're ugly and smell bad. Maybe that thinking "works" by some sort of human logic, but for all intents and purposes this AI is worthless.

[Inquisitor Recon]

I'm trying to think of the terrible things that would happen if you based this AI off my brain and gave it unprecedented processing power and resources at its disposal.

 

It would end very badly I think.

[Neoleviathan]

[metatheurgist]

We should never make anything that can think it's better than us. This includes children.

[mybudgee]

I'm trying to think of the terrible things that would happen if you based this AI off my brain and gave it unprecedented processing power and resources at its disposal.

 

It would end very badly I think.

[Lotion Soronarr]

We had one of the worlds foremost expert on AI give a speech on our college back in the day (that was years ago)

 

True AI is a pipe dream. We aren't even making baby steps, and since a brain is a non-deterministic, chaotic system, truly replaicating it seems impossible.

[Gravisanimi]

We had one of the worlds foremost expert on AI give a speech on our college back in the day (that was years ago)

 

True AI is a pipe dream. We aren't even making baby steps, and since a brain is a non-deterministic, chaotic system, truly replicating it seems impossible.

 

Those things do not exist.

 

We just don't know why it does what it does.

 

Or the how.

 

Personally, I'd use human brain tissue instead of silicon.

[Inquisitor Recon]

When you're mining for ore a mile under the earth's crusts for your robot masters think of CmdShep's warning.

[Gravisanimi]

Actually, I won't have you mining a mile under the crust.

 

Humans can't survive well at those depths.

 

In fact, computers don't work well in Earth-like environments.

 

If I were a super-computer out to destroy humanity, I'd set out to make my perfect habitat. I'd make this rock barren of all water, all life, an rock devoid of features and dust, with no atmosphere.

 

Incidentally, this would kill all humans.

[Eternal Phoenix]

They'll steal our women!

 

[Lotion Soronarr]

They have gone too far!

 

[Isichar]

Every night I sacrifice an external harddrive to our future overlords and masters.

 

I'll be the one laughing come judgement day.