Will human's still be relevant in 2500?

AnTiFreeze3

Humanity isn't stupid enough to allow robots to become more intelligent than we are. There is a huge level of unpredictability that follows something like the Singularity (not black-holes).

I don't see why we can't just cut them off at ape-like intelligence and force the robots to do manual labor for us.

AnTiFreeze3

Number Nine

"Cut them off"? How? Pass a law against it? A law specifying that all research in computational neuroscience and artificial intelligence must stop as soon as our technology reaches a point at which AI is quantifiably identical to ape intelligence? No.

Ryan_m_b

Why not? Politics and legal processes have halted scientific development in fields that will be far less controversial than this. Just think of human cloning. I'm not swayed by the "if its possible someone will do it" argument by itself.

Ryan_m_b

I have to ask, relevant to what?
It's evident that all life is not impossible and I have no idea what you mean by "overcomplex". By what reasoning did you come to that conclusion? And what do you mean "contribute". It seems like you're talking in terms of trophic flow but I can't imagine what you are getting at.
Be wary about the narrative of progress. There is nothing set in stone that human civilisations will always continue to become technologically more advanced, the truth of the matter is that we have no precedents for high energy civilisations and any speculation over the future (at these timescales and concerning this type of technological change) however well reasoned can't be more than speculation.
It means that we stand on the shoulders of giants.
You can't assert that the technology exists in one breath then insist it's a matter of time in the other. Firstly it doesn't exist and secondly strong AI/AGI has been coming real soon for decades. Whilst there has been progress in making software that acts intelligently it's not apparent to me that we've made strides towards strong AI. Reason being that general intelligence is ill defined in humans and its not even clear what it is. If we're going to try and create it from scratch or reverse engineer the brain (and they are both tasks of Herculean difficulty) we'll have to address the issue of what intelligence is along the way.
You're an AI researcher working to produce strong AI?
Superfluous to what? We're already superfluous to a lot of things and not to others. What changes if you propose strong AI?
My vote is that the premises of this discussion need reevaluating, see my previous post for more on this.

Number Nine

I didn't mean that it couldn't be done, only that it would be nearly unenforceable. Human cloning requires tremendous and specialized resources, but I can write a biophysically realistic model of the basal ganglia that will exhibit reinforcement learning in my bedroom. This will only become easier as computing power increases. Banning intelligent computer programs would be only slightly less difficult than banning cryptography. What's more, how would you quantify the level of intelligence beyond which computer programs would be prohibited?

The government would necessarily carve out exceptions for itself (likely for military purposes), which means that the necessary technology and knowledge would almost certainly be developed. It would be impossible to keep that knowledge from becoming public; they tried the same thing with public key cryptography and failed miserably.

Ryan_m_b

I agree and disagree. I see your point but writing the program for strong AI is likely to be insanely difficult, require very specialised training in AI science and possibly even require expensive and specialised equipment (the latter may change if Moore's law continues to the point where domestic/commercial machines match the requirements for strong AI, whatever that may be...). However I suppose it's possible that once the work is done the data could be copied and pasted.

Either way I maintain the topic is not as simple as "could be done = will be done".

AnTiFreeze3

Ah, and I suppose you were a strong advocate for advancing the research of nuclear weapons? Just because we are capable of advancing a field does not mean that we should do it. Like I said, there is a large amount of unpredictability, at least with our current understanding, as to what would happen if we allowed an AI smarter than we are to exist.

mfb

I prefer this world over a hypothetical one where North Korea has nuclear weapons and other countries have not.
Right, but the opposite is true as well: We should not stop all research we can do.
And there is Edit: source being discussed how to make an AI friendly.

AnTiFreeze3

I despise hypothetical arguments, but I'll try to humor you:

Theorize all you want, but there are virtually no circumstances under which the spending of trillions upon trillions of dollars on apocalyptic weapons could be considered a noble endeavor. And considering the primacy of North Korea and the intellectuals in its possession, I highly doubt they would have been able to produce any nuclear weapons of their own without outside influence; I would like to remind you that they have only recently come upon six pitiful nuclear weapons, and have done so over sixty years after the U.S. made the first nuclear weapon. And I can assure you that they did not reach this point on their own.

And my original statement was meant to be an example of how we should not strive to know everything about something simply because we can. Besides, the best way to hinder any self-extinction of humanity is to simply never develop the technology to kill ourselves off in the first place.

I don't recall ever promoting that all research should be stopped. All I have ever said is that, when certain research could be considered a threat to humanity, we ought to severely think about what we're doing before we make any wrong decisions. Blindly striving forward without thinking about the consequences of what we may come to know is a decadence at the very least.

The edit does make it a little difficult to understand, but I'm assuming that you meant to include a source stating that there is research being done on how to make potential AI friendly. I would be all for this; I thought about including a statement at the end of my last post stating how, if research were done that could guarantee the safety of this pursuit, then I would be completely pleased.

Ryan_m_b

Sorry but that source is not acceptable. I'd like to remind members to post links to credible sources I.e. peer review papers from credible AI journals. I doubt I need to remind anyone that this is a topic that had been commented on voraciously by various bias organisations that are not suitable sources for a PF discussion.

mfb

Of course. I think we all agree on that point.

@Ryan_m_b: Sorry. It was not my intent to use it as source of actual research.

@AnTiFreeze3: You can check the wikipedia article for a list of publications (see the references there).

Number Nine

You didn't respond to my question, which had to do with pragmatics and not morality. Nuclear weapons require specialized equipment; programming does not. How do you intend to enforce your proposed ban?

2112rush2112

Oh my gawd in jeebus I can't believe what I'm reading.
This forum (General Discussion forum) isn't the Quantum Mechanics forum, Evo! Telling this guy that there are "rules against overly speculative posts" in this forum is like telling your students on the campus green during lunch that there shall be no discussions about the possibility of interstellar life. "Enough of this balderdash talk! Not on my campus, no sir!"

Evo

The rules in General Discussion are the same as in the rest of the forum. We allow humor, personal interest discussions, etc... but only as long as they are within the forum guidelines.

micromass

If you know where that picture is from, then I got to say that you have great taste in music.

DiracPool

YOU! Yes YOU!!! ...the Lad recons himself a poet...and probably a physcist as well. Absolute RUBBISH. Get back with your work!