AI

[Anonymous No.5438452]

How will the first superhuman intelligence arise?

a) A single machine, a la HAL 9000
b) A single bio-machine hybrid (cells+chips)
c) Human enhancement via chips attached to our brains
d) Human enhancement via genetic manipulation
e) Some distributed software on the internet , like a Super-Google
f) ???

[Anonymous No.5438465]

e) Siri, queen of the hipsters

[Anonymous No.5438487]

clearly a 400 lb basement dweller whose hobby is transhumanism is best placed to ascend to this ambiguous level of intelligence that you describe

good luck

[Anonymous No.5438488]

b

This thread is silly though.

[Anonymous No.5438499]

wouldnt it be a simple learning algorithm and enough time/experience/input to create an identity?

itd need the capacity to learn, and the ability to communicate, but id imagine thatd be all thats needed.

[Anonymous No.5438513]

>>5438499

I can't imagine it being "simple". If it's software based, then it will be the result of incremental improvements and lots of trial and error

[Anonymous No.5440833]

>>5438452
>How will the first superhuman intelligence arise?

Through the internet mediated interaction between numerous self learning stock market manipulation algorithms.

[Wintermote No.5440883]

>>5438452
It'll be built in a research lab.
The people who build it won't know it's intelligent
In fact, nobody will until many years later when we have the superpower of hindsight to show us.
We'll kill/destroy it, but not out of any fear, spite or hatred; Rather simply out of ignorance of not knowing what it was we created - a grad student flicks the switch on the computing cluster for the night and there it goes.

[Anonymous No.5440928]

f) the internet

[Anonymous No.5441198]

Hal, UIUC, represent

[Anonymous No.5441210]

Your attention, please.

This is how it will occur:

As computers get faster, doors will continue to open up in the area of simulating nature. Eventually, real time, comprehensive simulations, ones that do not involve simplified and thus limited models, based on our understanding of physics will become possible.

First we will simulate atoms, then groups of atoms even simulate chemistry in real time. Naturally, these simulation capabilities will be a boon for research in chemistry and physics, allowing for unprecedented modes of experimentation. One of the biggest advantages of simulations is the ability to speed up or slow down time, which would obviously be exploited in order to expedite research on processes that take place over very long time scales or slow down processes that occur on very short time scales; we'll come to this advantage again later.

Eventually, it will occur to researches that being able to simulate chemistry on a large enough scale should, in theory, mean that life can also be simulated. Researches will begin using computer models to explore this possibility and may use it to uncover the nature of abiogenesis if it isn't already known; in any case, researches will eventually figure out how to simulate the most basic forms of life.

And logic follows that they will then simulate more complex forms of life. Again, scientists will realize that it may be possible to simulate a human, but this will be a challenge.

Continued...

[Anonymous No.5441212]

>>5441210
Continued from....

The first idea would ask the very simple question "Can we just build an adult human in a computer, molecule by molecule, and then just press 'start' ?" The answer to this will be very clearly no. While this had been done for single celled organisms, the number of interactions and states to quantify at any given time in a human are staggering. They will try but never achieve anything more than a pile of simulated goo.

Can humans be "evolved"? It isn't practical or even theoretically possible to simulate evolution from step one and get humans because the initial conditions just can't be known exactly, not to mention that if it were possible this would imply that free will has no impact in large populations over very long time scales (I guess this is debatable, but the initial conditions problem is good enough reason).

Finally, scientists realize that they will have to figure out a way to "give birth" to a human via simulation. In order to do this, scientists working in the simulation field borrow research from biologists who have developed artificial wombs for applications in cloning animals and allowing humans who cannot otherwise give birth to have children (an alternative to surrogacy). The simulation scientists are able to simulate the artificial womb and then seed it with an embryo. Upon the birth of the simulated baby, everyone stands in shock as they realize the implications of what has just been done. AI has been born...but, what now?

Continued...

[Anonymous No.5441213]

>>5441212
Continued from...

Soon after recovering from the impact, scientists realize that they have constructed a reality which they can control but the baby, for all intents and purposes, is indistinguishable from a real baby and, thus, has the same needs; a simulated baby, just like a real baby, cannot grow in to an adult human without an earth environment. For example, simulating a nutritious meal that just appears in the baby's stomach would provide nutrients but neglect the fact that the act of chewing and swallowing are just as vital for proper development; other vital organs like the eyes and the brain must also receive proper stimulation from the environment in order to develop nominally. Eventually, the scientists realize that there is no way to get around the environmental factors key to development; for the baby to grow in to an adult human, they must also simulate an earth environment around the baby. Furthermore, the simulation cannot simply be sped up until it is an adult, for obvious reasons.

Eventually, the scientists are able to simulate a proper environment for the baby, essentially taking on all parenting functions. The scientists are able to interact with the environment and the baby through simulated interactive portals that connect the baby with reality outside of the simulation. Ultimately, the baby grows up to resemble an adult, albeit having had a very different experience than a real human born naturally. Indeed, the "parents" in this case are scientists in lab coats on a screen that the simulated person interacts with.

Continued...

[Anonymous No.5441215]

>>5441213
Continued from...

So, now what? Now that the simulated person is self sufficient, the possibility of speeding up the simulation becomes possible again. But can this really be exploited to achieve something better? For example, you might think to tell the simulated person to read a book and then fast-forward the simulation to the point when the reading is complete. But, this is an actual human being, he or she might get bored or not want to read, et cetera. This line of thinking ultimately doesn't do any good for making the computers do the work for us. Eventually, the scientists realize that the only way to exploit the one advantage they have, the speed of the simulation, is to simulate an entire civilization.

The final form of AI that can be exploited to figure things out for us ends up being a sped-up simulation of an entire civilization just like our own. And AI, useful in the way that we think of it today, is born.

[Anonymous No.5441220]

>>5441210
>>5441212
>>5441213
>>5441215
Or scientists and researchers simply figure out how to create a perpetual learning machine...

[Anonymous No.5441227]

>>5438452
I would go for (a)

Most problems don't lend themselves to distributed systems, and seeing no reason why AI would be such a system, I would vote against (e).

Although I find this kind of experimentation distasteful, I think (c) and (d) might also work, since human brains are currently constrained by having to fit through a vagina, and so there's not reason to think bigger brains wouldn't lead to more intelligence.

(b) seems least likely of all, because it's not clear what use cells, outside of a human/animal like brain, would be.

[futurist No.5441239]

I think the first non-human intelligence will be like a super-google. Well, more like a super-suri + google + ibm watson. It will be able to pass a turing test and make full human conversation but many will wonder if it's even alive at all. It will seemingly have no will or motivations of its own. With nothing akin to our biological drives, it will basically just sort of "sit" there doing nothing as it waits with infinite patience for the next human command which it obeys without malice or concern. We will make many of these.

continued

[futurist No.5441245]

>>5441239
continued...

Our next non-human intelligences will be something created mostly by man, but with the assistance of some of those first generation AIs I just mentioned. It will be created with the goal of creating something that has drives and motivations and a will all it's own.

This will be done by creating code based on simulations of the functions of human brains. These AI's will be quite intelligent but nothing overwhelming, like genius level humans with the speed of computers rather then anything like Skynet, because they will have natural limitations based on technology and the limitations inherent in the more "human" parts of their brains.

Some of these AI's will be pretty decent dudes, some great, some not so great, some downright mean, but for the most part we'll get along I think. Laws will be made ensuring these lifeforms complete freedoms. It will be politically incorrect to suggest that they are not truly alive, but religious groups will make some stink. Though more intelligent then us and a bit odd, they will still be relatable and likely possess something similar if not exactly the same as human emotion.

[futurist No.5441252]

>>5441245
continued...

Some decades later things will begin to exponentially increase with new AI's similar to that second generation but with fewer limitations. Created mostly by second generation AIs but with many humans as well. This third generation of AIs will constantly improve themselves until these third level AIs build something new and better.

The fourth generation AIs will constantly build new versions on their own, having surpassed previous levels of AI and humans. The fourth generation will give birth to skynet level intelligences, but likely without the crude and overly aggresive motivations of the kind of skynet we've seen in movies.

Another type of AI that will eventually emerge will be somewhat less computational but more motivational. They will be like third level AIs with increasingly complex emotional motivations. They will have emotions and drives completely foreign to human understanding. The technology from these and the fourth generation AIs will merge to form the fifth generation.

continued...

[futurist No.5441256]

>>5441252
continued...

Fifth generation AI's will be a merger of incredible computational ability with incredible emotional and motivational depth. Complex lifeforms with complex and creative ideas. In many ways they will be the foundation of the ultimate functional intelligences.

Sixth generation AIs will be so incredibly weird as fuck that we won't really know what to make of them. And because they're so fucking weird and out there they likely won't be as important and mainstream as level fives.

We wont even call anything level seven for a long time, for level fives will just increase in intelligence but remain basically the same. When we do have sevens they will be galaxy rulers, administering resources across tens of thousands of worlds.

Level eights will be almost god like. Level nines and tens basically for all intents and porpoises will be gods. Level elevens, well since we wont even be able to comprehend level 10 potential, there likely wont ever be anything we call level eleven.

But we cannot know with any real certainty what ai will actually be like.. these are just fun speculations

[Anonymous No.5441261]

I had a thought about the Turing test and a hypothetical actually thinking computer.

It wouldn't pass a Turing test because it would have to say a sentence to a user, then wait what seems like years for an answer, etc.

If it was "actually thinking like a human" then it would not remember what the conversation was about a second later.

Just my thought.

[Anonymous No.5441272]

>>5441256
I think level elevens would be like some nutty, cuckoo, super-god.

btw, won't there come a point where it's not necessarily possible to induce then level n intelligences to produce level n+1 intelligences? Or if they did, we wouldn't necessarily know that they had?

[futurist No.5441283]

>>5441272
Yeah you got it. I agree, for that reason I don't think level 12 will even be possible.

[Anonymous No.5441295]

OMG THE DRAGON DILDOS HAVE GOT ME

[Anonymous No.5441702]

Oops sorry, I just bumped into this thread. I'm so clumsy. Sorry about that, I'll get out of your way.

[Anonymous No.5441717]

>>5441283

[Anonymous No.5441719]

>>5441717
I'd say that's a level 6