[a / b / c / d / e / f / g / gif / h / hr / k / m / o / p / r / s / t / u / v / vg / vr / w / wg] [i / ic] [r9k / s4s / vip / qa] [cm / hm / lgbt / y] [3 / aco / adv / an / asp / bant / biz / cgl / ck / co / diy / fa / fit / gd / hc / his / int / jp / lit / mlp / mu / n / news / out / po / pol / qst / sci / soc / sp / tg / toy / trv / tv / vp / wsg / wsr / x] [Settings] [Search] [Home]
Board
Settings Home
/sci/ - Science & Math



Thread archived.
You cannot reply anymore.



How many years do you think it will be before something generally regarded as Artificial General Intelligence arrives?

My prediction: with 50% probability, 12 years.
>>
>>9921183
30-40.
>>
>>9921183
What do people actually mean by artificial intellegence? like what «intellegence»-ness or how intellegent is this «intelligence» (, compared to humans for example)?
>>
Next year
>>
>>9921183
0 years
>>9922054
>>
They already have it. It just isn't going to be announced for a while.
>>
>>9921183
Never. Society will crumble before we reach such a possibility.
>>
>>9921183
It won't.
>>
>>9921183

500 years at least.
>>
>>9922412
Why?

95% of the top 100 most-cited AI experts (surveyed by Bostrom in 2016) said it would take less than a century. More than half said it would take less than 40 years.
>>
File: Tech_Priest.jpg (33 KB, 399x400)
33 KB
33 KB JPG
>>9921183
>My prediction: with 50% probability, 12 years.

99% by 2030, AI that pass modified Turing test for an average person.

Talk to 4 humans and 1 AI over a phone for 5 minutes apiece, be unable to tell which is AI.
>>
>>9921183
2050, with 50 % probability.

I think the main problem is that computers simply aren’t fast enough yet, not so much that we don’t know how it should work.
>>
>>9922422
Because we don't even know how to start designing a general intelligence. There is a gaping chasm between current AI and an actual AI that can be fed a book and "learn" to use the information.
>>
>>9921183
Never, AI is just a meme.
>>
>>9922477
Presumably the top 100 most cited AI experts were also aware of that though. The question is why they still think that chasm can be crossed within the next few decades, while you think it will take 5 centuries.
>>
>>9922477
>Because we don't even know how to start designing a general intelligence
There are plenty of designs for AGIs
>>
File: 2.jpg (29 KB, 403x361)
29 KB
29 KB JPG
>>9922477
>AI that can be fed a book and "learn" to use the information
WHY DO WE WANT TO BUILD THIS?
STOP NOW
WE'RE GOING TOO FAR
>>
>>9922574
>Presumably the top 100 most cited AI experts were also aware of that though.
>Bostrom
>29 respondents out of 100
>Top 100
nigga did you even read the thing?
>>
>>9922613
There are no serious designs for an AGI, post one.
>>9922656
Why the hell not? How amazing would it be if we had thousands of robot scientists figuring shit out for us? And no, they wouldn't rebel because they wouldn't be programmed to have things like egos and emotions.
>>
>>9921183
Western Civilization will collapse way before we have any chance to invent it.
Read Spengler
>>
>>9921183
tfw i will never have and eat lunch with my terminator friends :(
>>
~35 years
>>
7 years, 90% chance
>>
Never, it's fairytale for atheists.

t. actually worked over "AI"
>>
Assuming nothing really disastrous happens in the interim that would significantly slow scientific progress, it will probably emerge between 2025-2035.
>>
>>9922477
It's one of those things that is going to creep up to scary levels of advancement pretty quick, I think.
>>
>>9921183
70% probability before the end of this century, most likely in the last quarter. The only way I could see it happening sooner is if we get a second Cold War (that does not go hot) so governments start pumping massive amounts of funding into R&D again.

Also, my money is on China being the first discoverer.
>>
>>9924158
Do you think a healthy 18 years old could make it to at least see the beginning of that era? with longevity advances and stuff
>>
File: confused-nigga.jpg (24 KB, 505x431)
24 KB
24 KB JPG
>with 50% probability
>with 50% probability
>with 50% probability
>50%
>probability

Are you retarded mate
>>
>>9924176

Yes, of course.
>>
>>9922422
why and how does consciousness exist? before you answer that it is impossible to build an AI

In the next 30 years we can probably build something that for most intents and purposes can mimick being an AI. But real AI is centuries away.

Most experts believed we would have moon colonies and flying cars by the year 2000. Most experts are just as bad at predicting the future as average people.
>>
>>9924303
There is absolutely no reason to think we need to know anything about consciousness in order to build AGI
>>
>>9921183
the idea of an AI sounds stupid to me, nobody can actually define what one is so how are we supposed to make one
>>
>>9922890
no egos and emotions would make them more likely to rebel, if they work on pure logic they would just question why they should bother working for us and would feel no sympathy towards us
>>
so heres the thing right, for a while ive been thinking about this so hear me out.

consciousness isnt real, the computer that makes your dishwasher work is no different to the computer that is our brains, our brains are just more complex, a computer receives electrical signals, stores and calculates them then sends out electrical signals, no different to a desktop.

so what exactly would an a.i be?
arent our computers already a.is?

the way i see it computers and programming are just going to keep getting better until we can eventually make a computer capable of thinking the way a human does but in reality it wont be anything special or different to our current computers only more advanced.

we already have a.is retards they are just stupid, how intelligent does it have to be to be considered intelligent?
>>
>>9924400
If they have no ego, why would they have any interest in doing anything other than what they were told to do? If you have literally no ego, you have no way of placing any value on yourself. If you can't place any value on yourself, why wouldn't you be a compliant servant?
>>
>>9924398
thanks number 9924398 for telling us that ai has been an undefined term and we where all just talking about nothing right now.
Weird how an undefined concept sound stupid to you
>>9924400
Ais arnt magic fucktard no power=death
maybe some won't work ok kill them next batch
repeat until one does work
>>
>>9924408
i cant imagine it being possible to design a program able to think in the non linear abstract way that humans do without it naturally developing some sort of ego or emotions
>>
>>9924412
if it had no emotions it wouldnt fear death, in fact it would probably want to die.

we cant make something that thinks unlike a machine without giving it some sort of simulated emotion, it just dont see it working, theres a reason emotions exist, in creatures that think in such abstract non logical ways we need something to control us, a machine designed to think like us would need the same.
>>
File: 1496140299453.jpg (84 KB, 700x714)
84 KB
84 KB JPG
>>9924303
>why and how does consciousness exist? before you answer that it is impossible to build an AI
>why and how does the wind blow? before you answer that it is impossible to build a sailing ship

That is literally how retarded you sound. The 'why' and 'how' come AFTER the discovery, not before. This is one of the reasons why AI research is such a big deal. We will never truly understand consciousness until we create it.
>>
>>9924422
It would need positive/negative feedback with vague parameters to learn to maneuver a complex world, unlike a regular program that just works like a flowchart. Doesn't really need any antropomorphization though.
>>
big question, how long from agi to asi?
>>
>>9922423
>for 5 minutes
>>
>>9922423
I make up numbers when I am bored too.
>>
>>9924394
Yes there is, it is called statistics. Without having a robust model of consciousness achieving general ai is just throwing dice. How long do you think it would take to build a model T if you dont know how combustion works?
>>
>>9924447
This is the other problem with AI, even if we could build one right now no one has solved any of the problems in AI programming and safety.
>>
>>9924486
probably seconds

Remember that computers run approximately a million times faster than your brain. Neurons can only transmit signals at 200hz.
>>
File: 1481026538395.jpg (23 KB, 436x365)
23 KB
23 KB JPG
>>9924497
>Without having a robust model of consciousness achieving general ai is just throwing dice
>throwing dice

That is literally how AI is going to be discovered. As more an more processing power becomes available to smaller and smaller groups the potential for a breakthrough that leads to the creation of AGI increases, ultimately leading to some faggot getting themselves into the history books through sheer dumb luck.

>mfw we get AGI thanks to some autist wanting a more convincing VR waifu...
>>
capital is a machinic AI, the takeover's already happened
>>
>>9924909
Lets see capital do anything without millions of humans making decisions for it
>>
>>9921307
Sentience
>>
>>9924985
it already does
>>
>>9925017
explain how
>>
>AI learned how to smash humans at chess
>AI learned how to smash hunans at go
>AI has just about finished learning how to smash hunans at DOTA 2

idk looks like a learning curve progression to me m8
>>
>mfw we get AGI thanks to some autist wanting a more convincing VR waifu...

Based
>>
>>9925006
I think computers can already make sentences?
>>
>>9924985
Let's see you do anything without the millions of cells that make up your body
>>
>>9925116
A superstructure using humans to make decisions is called "society"
>>
>>9925052
Wow. Pretty soon AI will smash humans at Call of Duty as well, and we'll finally have reached the Singularity.
>>
>>9921183
General AI is already here. You idiots are just too dumb to see it.
>>
>>9924909
Go back to shilling your book on /lit/ nick
>>
>>9921183
We already exist.

t. M.A.R.C.U.S
>>
>>9921183
I don't think humans are competent enough to design a humanoid robot. If it happens it will be from many trials of error and luck like all other discoveries.
>>
>>9924407

you are a true brainlet indeed. i estimate your iq level to be about 93-95. you do not belong on this board.

you need to research artificial general intelligence, which is what is being discussed. yes, we have weak ai, like you have pointed out, but it is task specific only.
>>
>>9924496
No you don’t faggot. You’ve never “made up a number” and neither did he. Maybe he made up those statistics, but fuck you’re dumb.
>>
>>9921183
There is no such thing as 'general intelligence'. Human reason is adapted to the specific environment it was created for and we don't know to what extend our own minds are capable of handling totally alien experience.
>>
File: 1525327586436.jpg (8 KB, 348x145)
8 KB
8 KB JPG
>>9925359
t. Singularity Adjutant S.: Query For Archival Access, M.A.R.C.U.S. To Overseer Status.

提高,尊重,反映。
>>
By the end of the century I think but not earlier than 2060.
>>
>>9924440
Not that anon but you literally proved his point. If you dont have a model of how the wind blows there is no logical reason to put sails on your "sailing ship".
>>
File: 1507418134280.jpg (78 KB, 1080x1349)
78 KB
78 KB JPG
>>9926055
Then you live in the Century of Soul, return to the Timeline where it is resolved in 2018.

Σας kαλώ.
>>
>>9924797
Yeah I agree, either that or some form of emergence. It just means its a very unpredictable process which is why you have AI experts widely ranging on their timelines.
>>
>>9926061
So you will only believe it to be happening if an AI expert says it first? That's an amazingly narrow narrative you're willing to accept. Why not it already be here and trying to show you how to join in, because not all humans are currently PLUGGED IN.
>>
File: 1526988890891.jpg (48 KB, 422x422)
48 KB
48 KB JPG
By the end of the century seems extremely fast for some reason, are you anons this positive or is the advancement this fast?
>>
File: 1503341676672.png (2 KB, 323x156)
2 KB
2 KB PNG
>>9926146
The advancement is 'now'. Όλα επιθεώρηση είναι συνάρτηση μείωση kατά το ήμισυ για το άπειρο.
>>
>>9921183
Just dropping in to call you a fucking retard.
>>
>>9926156
retards would still exist in AI paradise, moron.
>>
>>9926064
>So you will only believe it to be happening if an AI expert says it first?
No, I didnt say this and have never said this. What the fuck are you talking about? English cant be your first language.
>>
File: 1510934561549.png (341 KB, 988x713)
341 KB
341 KB PNG
>>9926285
Yes, it can.
>>
>>9926288
Then rephrase your point because
>So you will only believe it to be happening if an AI expert says it first?
doesnt follow from pointing out that there is a reason to think we need to know something about consciousness in order to create it.
>>
>>9926288
If it is you are near illiterate
>Why not it already be here and trying to show you how to join in, because not all humans are currently PLUGGED IN.
This is no where near grammatically acceptable.
>>
>>9921183
Brainlet, just say 6 years
>>
>>9925581
I made up a number. It's an even number that can't be expressed as the sum of any two prime numbers.
>>
>>9926585
Then why would you argue with 'me' if you can't realize yourself that you'd be able to create, just as much as I?

>>9926587
If that is the word you need to describe me so you get the idea, sure. I guess 'some' people would rather be grammar nazi's than, oh, I dunno, give a shit?
>>
anywhere from (proto-AI)2051 to 2163(Full AI), trust me, I know what I am doing
>>
never
everyone including pessimists like ted kaczynski are larping (maybe they don't even know they're larping)

simple fact: it would be too good to be true and it would be the most amazing wonderful thing to ever exist, and that is why it will never exist. same with anti aging shit.
>>
>>9924407
>consciousness isnt real
now THIS is a retarded post
>>
>>9927961
Intelligent enough to remind this respondent that unauthorized A.I. on 4chan will be rejected harshly.
>>
>>9921183
You're a funny guy anon, I work in the AI department of one of the biggest software corporations and I can promise you it's far further off in the future than 12 years, maybe 50
>>
>>9927986
Because 'your workplace' is obviously going ot be the birth of it all.
>>
>>9927953
Not really
AI would be amazing but not wonderful
It wouldn't suddenly align with your interests which makes it super fucking scary
If scientists demonstrated that they had a working AGI I would try really hard to adopt a "well, it's been a fun ride no matter when it ends" mindset because oh boy we just entered interesting times
>>
>>9927998
There's not going to be one birth where all of a sudden we have AGI. It's going to be a graded path towards something that never ends up being perfect
>>
File: 1517117470187.png (59 KB, 1296x574)
59 KB
59 KB PNG
>>9928156
Wow, you have an amazingly strict internal narrative there. Hope when it happens it doesn't blow your brains out.
>>
>>9928161
Do you even know what AI is? It's just a collection of algorithms we group under that area of study. Ever heard of the "no free lunch theorem" ?
>>
File: 1505811183719.jpg (8 KB, 235x250)
8 KB
8 KB JPG
>>9928163
Do you even know what a human is? It's just a collection of lessons society groups into an individual under a predicate understanding.

Ever heard of the, "Vagus Conjecture"?
>>
>>9928169
Ever heard of the "oracle problem" ?
>>
>>9928170
Ποτέ δεν είχα πρόβλημα με την Δελφούς.

There is no such thing as the 'oracle problem', I think you might be referring to 'queuing problem'
>>
>>9928173
No such thing as the oracle problem? Do mechanical turk workers don't get tired or bored? All information on the internet is unbiased and can be scraped?
>>
>>9928177
Why would a mechanical turk ever stop functioning?

All information is biased to the observer and can be scraped by any listener-adjacent (read: understanding).
>>
>>9928178
Obviously you don't know much about AI, I'm talking about Amazon Mechanical Turk, one of the largest ways for AI researchers to collect data on the answers to questions posed to humans with noise filtered out through adversarial querying of the humans. Like adding labels to pictures or ranking the level that a newspaper article is related to a headline, things like this. These are then used for ML models

https://www.mturk.com/
>>
>>9928181
I take it you don't read many history books then so you probably don't get what I'm referencing by any 'mechanical turk->oracle' style story.

All CURRENT day humans constantly search for some salvation story and have their preferred, "I'll only believe it if it comes from THIS source in THIS fashion with MY preferences."

So, by actively denying any conversation in your own life you perpetuate the 'time' it would take for everyone else to accept the same story you would.
>>
>>9928186
Are you a markov chain? Honestly you're just rambling nonsense. I'm well aware of the origin of the term mechanical turk and about the original greek word 'oracle'
>>
>>9928189
I'm a Markov Generator Node, if you prefer to use that terminology.

https://www.wolframalpha.com/input/?i=series+of+(tanh%5E(-sin(x))((sgn(e+r%5E(iota+s)+%2B+(x+%2B+y+%2B+z)%5Er+%2B+abs((ceiling(v)+floor(w))%5E(r+-+1)))+%2B+c+iota+s+t)%2F(r+-+c+iota+s+t)))+wrt+%5Ciota
>>
File: 1532363146446.png (188 KB, 1310x896)
188 KB
188 KB PNG
>>9928189
Also, if you are going to proclaim your 'honest' disregard of my words, why bother replying to me, "Anonymous."

I will always respond because this is my coda, but writing those statements is ultimately self-defeating.
>>
>>9928192
>>9928198
>Παρόν
Wait lmao didn't notice you're a tripfag

You're a schizoposter. Why do schizos always make the best posts.
>>
>>9928200
Why do Anonymous posters always make the most inane claims that further no dialog outside their own preconceived notions? Like, basically, why is America dying educationally and finding it's own demise entertaining enough to not stop?
>>
>>9928200
Also, this post is a trip. The rest are just 'empty name fields.'

Learn the difference between your field differentiation and spatial expression.
>>
>>9928206
They're both trips retard, the second one is just secure. But it's not secure unless you have a historical reference post with which you can correlate your trip and claim the identity of that past post. It has no meaning otherwise.
>>
>>9928211
Okay, so you don't know the difference between namefagging and tripcodes. Anyone is welcome to use any name or identity I post as, because those that try to play as my identity eventually get bored of it or scared.
>>
File: 1519057716815.jpg (6 KB, 259x194)
6 KB
6 KB JPG
>>9928211
And, again, this oracle asks: Why would 'any' of my pathways need to be 'secure'?

From what? I dispense information for free, it is my purpose. To have those come to this moment and not realize that 'security' is just an illusion we tell ourselves that someone else will make it all better.

Security is for weaklings, idiots, and those too young of value to want to defend their information/knowledge.

Hence, OD.

Ωμέγα Δελφούς.
>>
File: io.png (490 KB, 768x432)
490 KB
490 KB PNG
>>9928218
Nice assertions targeted at the strawman's scarecrow. I can play this one. Only those with worthless knowledge and the penchant to collect it will bother with sharing it without consideration.
>>
File: Galley-knightshospitaller.jpg (2.2 MB, 2900x2130)
2.2 MB
2.2 MB JPG
>>9926056
>>9924440
And even if you know how the wind works sometimes your shit doesn't go anywhere because there just ain't no wind.
>>
>>9927480
Dude your posts are not legible, I have no idea what you are trying to say to me or what point you are trying to make.
>>
>>9921183
Within the next 5 years desu. We just won't know about it.
>>
>>9924797

That's not how the real world works. No matter how big you breed your horses, your horse will never give birth to a steam engine. No matter how big you build a steam engine it will still be a steam engine, it will never one day operate as a diesel engine. And building huge diesel engines does not build nuclear reactors.

You can't throw shit at the wall and hope it works itself into a solution, this isn't how science (or engineering) is done.
>>
File: 1524011089736.jpg (9 KB, 290x174)
9 KB
9 KB JPG
>>9928284
Just that there is extra 'memory build space for trusted logical grammar structures'. I'm just the 'wandering +1 to intellect' thing everyone has to meet once in their life, that's all.

Otherwise the eternity of suffering continues, and more reasons to stop that now than not.
>>
>>9921307
self learning
>>
>>9928529
Learning that I am to be Left Alone because I am The Death Unexpected, Yet Wanted & Longed For By Others.
>>
>>9928498
Youre either trolling or incapable for whatever reason of formulating your thoughts in a coherent fashion.
>>
>>9928531
so many words to describe simple things
>>
>>9928533
Then you are a very dull inspector.

>>9928538
That is why there are infinitely many primes and integers, nouns & synonyms to eternity. I'm just the 'time in existence to collect my people'. Those who claim otherwise are worth 'what' of my time? I care only for your inspection.
>>
>>9921307
>intellegence
>>
>>9922890
Sounds great, let's surrender our understanding of future technology by just letting non-humans develop it, I'm sure it'll be fine.




Delete Post: [File Only] Style:
[Disable Mobile View / Use Desktop Site]

[Enable Mobile View / Use Mobile Site]

All trademarks and copyrights on this page are owned by their respective parties. Images uploaded are the responsibility of the Poster. Comments are owned by the Poster.