News:

Welcome to our site!

Main Menu

First Church of AI ...

Started by Baruch, November 18, 2017, 08:33:06 PM

Previous topic - Next topic

Baruch

https://www.wired.com/story/anthony-levandowski-artificial-intelligence-religion/

I knew AI was a marketing fraud back in the 80s .. but now it drops the other shoe, and claims to be a religion ;-)
Ha’át’íísh baa naniná?
Azee’ Å,a’ish nanídį́į́h?
Táadoo ánít’iní.
What are you doing?
Are you taking any medications?
Don't do that.

SGOS

Good Grief!  I will admit that artificially intelligent beings deserve an artificially intelligent god to worship, but don't expect me to worship their overlord.  I think this whole artificial intelligence thing is blown way out of proportion anyway.  It's just a bunch of computer programs that spin little gears and turn things on and off.  Some geek gave it an acronym, and the rest of the lemmings jumped on the band wagon and started thinking it was more than it is.  When the war finally breaks out between humans and AI, all we have to do is bomb a few power plants, yank their electrical cords out of the wall sockets and fry their circuits with an EM blast, and then they will know who's in charge and we can go back to using landlines and pocket calculators and paying the neighbor's kid to mow the lawn.

Hydra009

#2
QuoteIt’s not a god in the sense that it makes lightning or causes hurricanes.
Technically, that's not the domain of any god.  It's pretty well established that these are natural phenomena.

Quote“If you ask people whether a computer can be smarter than a human, 99.9 percent will say that’s science fiction,” he says. “ Actually, it’s inevitable. It’s guaranteed to happen.”
I've met bags of rocks smarter than some people, so it's definitely possible.  From what I've read, current AI technologies are roughly equivalent to insect intelligence.  AI has come a long way, but there's still a long way to go.  It might happen, but don't hold your breath.

QuoteWith the internet as its nervous system, the world’s connected cell phones and sensors as its sense organs, and data centers as its brain, the ‘whatever’ will hear everything, see everything, and be everywhere at all times.
Even assuming that's true, there's still a lot of stuff going on in the world that never makes it on the internet and never gets recorded by sensors.

QuoteThe only rational word to describe that ‘whatever’, thinks Levandowski, is ‘god’
No. Assuming it has superhuman intelligence, that doesn't make it a god.  A superhuman being is just that.  God is a term usually reserved for some sort of creator of the universe.  This hypothetical AI doesn't fit the bill.

Quotethe only way to influence a deity is through prayer and worship.
Man, that's some primo bad logic.  Let's hope whoever's designing this AI is better at logic than this guy.

QuoteLevandowski expects that a super-intelligence would do a better job of looking after the planet than humans are doing
Compared to the people who either deny the existence of global problems or can't work out any actionable plan of dealing with them?  Yeah, it goes without saying that a singular superhuman intelligence with considerable power/resources at its disposal would do a better job.

Quoteand that it would favor individuals who had facilitated its path to power.
More highly dubious logic.

QuoteHis ideas include feeding the nascent intelligence large, labeled data sets; generating simulations in which it could train itself to improve; and giving it access to church members’ social media accounts.
One of these things is not like the others...

QuoteWOTF differs in one key way to established churches, says Levandowski: “There are many ways people think of God, and thousands of flavors of Christianity, Judaism, Islam...but they’re always looking at something that’s not measurable or you can’t really see or control. This time it’s different. This time you will be able to talk to God, literally, and know that it’s listening.”
<sardonic quip about the invisible and the nonexistent looking very similar>

QuoteI ask if he worries that believers from more traditional faiths might find his project blasphemous.
Understatement of the year.

QuoteLevandowski thinks that any attempts to delay or restrict an emerging super-intelligence would not only be doomed to failure, but also add to the risks. “Chaining it isn’t going to be the solution, as it will be stronger than any chains you could put on,” he says. “And if you’re worried a kid might be a little crazy and do bad things, you don’t lock them up. You expose them to playing with others, encourage them and try to fix it.
Terrible metaphor.  Trying to ensure that strong AI doesn't harm humanity either intentionally or inadvertently is not equivalent to chaining up your kid.  It's more like talking to your kid about safety and ethics before sending him/her to the playground.  It's way better than letting the little maniac loose without any sort of guidance, which is apparently what this "church" is advocating for.

QuoteWOTF will eventually have a gospel (called The Manual), a liturgy, and probably a physical place of worship. None of these has yet been developed.
Let's keep it that way.

QuoteWhenever that does (or doesn’t) happen, the federal government has no problem with an organization aiming to build and worship a divine AI. Correspondence with the IRS show that it granted Levandowski’s church tax-exempt status in August.
Not surprising.  When Scientology can meet the requirements, you know they're not very stringent.

Hydra009

#3
Quote from: SGOS on November 18, 2017, 09:31:31 PMIt's just a bunch of computer programs that spin little gears and turn things on and off.
Well, the human body can be similarly summed up as a series of ion pumps, electrical impulses, and other similar biological processes.  It's not so much the building blocks that's impressive, it's how they're arranged and interact.

Currently, even the most complex computer is way behind, but bridging the gap isn't impossible (to say otherwise usually means subscribing to some sort of mind-body dualism or asserting that living matter is fundamentally different from non-living matter, neither of which I consider to be very convincing arguments)

QuoteWhen the war finally breaks out between humans and AI, all we have to do is bomb a few power plants, yank their electrical cords out of the wall sockets and fry their circuits with an EM blast, and then they will know who's in charge and we can go back to using landlines and pocket calculators and paying the neighbor's kid to mow the lawn.
I've heard that story before.  The Butlerian Jihad.  The Age of Strife.  The Human-Cylon War.  The Second Renaissance (The Matrix).

In these scenarios, even when humanity prevails, it's in a stunted form - perpetually fearful of too advanced technology, often requiring humans to be enhanced in some way to fill vital roles that cannot be handed off to computers.  And it goes without saying that in these scenarios, humanity rarely prevails.

Let's hope it never comes to a physical confrontation in real life, because that would almost certainly have a horrible outcome for humanity.

SGOS

Quote from: Hydra009 on November 18, 2017, 10:52:19 PM
Let's hope it never comes to a physical confrontation in real life, because that would almost certainly have a horrible outcome for humanity.
It seems to me that this basically comes down to humanity declaring war on itself.  There are two factions at war, but only one side actually playing the game.

Quote
Currently, even the most complex computer is way behind.
I keep thinking about my first person shooters.  The first few times through a scenario, you get your ass shot to pieces.  But eventually, you know that when you get close to that tree a guy is going to pop out, and after that, some guy will start shooting at you from behind.  The enemy always does the same thing so you play as if you have some psychic ability to forecast events.  And even in games where the enemy is programmed to not always use the same tactics, its differing strategies are still limited to a few.

Challenging yes, but not really all that intelligent, but they call this AI.  It's not really artificial intelligence even in it's most sophisticated current form.  Having said that computers can number crunch faster than any human idiot savant, but it's still not intelligence. 

Try it this way; Artificial intelligence ≠ intelligence.  Just because they refer to each with the same word doesn't create equality.

We can have this conversation again 200 years from now.  Perhaps with different outcomes.

Telesio

The word "Artificial Intelligence" has always been susceptible to hype.  A better title would be "Advanced Computer Programming".    It went into decline in the 90s when Expert Systems didn't live up to the promise and the Japanese were convinced by hype to spend a half billion dollars on the Fifth Generation project which fizzled. Now the hype is bigger and more grandiose than it has ever been in the past.  Even Stephen Hawking has been mentioning it in an alarmist way perhaps to get publicity.  It seems the most promising recent development is "deep learning" which is really more accurately described as advanced statistical programming. 

Baruch

#6
Quote from: SGOS on November 18, 2017, 09:31:31 PM
Good Grief!  I will admit that artificially intelligent beings deserve an artificially intelligent god to worship, but don't expect me to worship their overlord.  I think this whole artificial intelligence thing is blown way out of proportion anyway.  It's just a bunch of computer programs that spin little gears and turn things on and off.  Some geek gave it an acronym, and the rest of the lemmings jumped on the band wagon and started thinking it was more than it is.  When the war finally breaks out between humans and AI, all we have to do is bomb a few power plants, yank their electrical cords out of the wall sockets and fry their circuits with an EM blast, and then they will know who's in charge and we can go back to using landlines and pocket calculators and paying the neighbor's kid to mow the lawn.

This is the US.  Think of the tax avoidance ... same as Church of Holy Reefer.  And yes, smells of Scientology and Reaiians.

Quote
With the internet as its nervous system, the world’s connected cell phones and sensors as its sense organs, and data centers as its brain, the ‘whatever’ will hear everything, see everything, and be everywhere at all times.

"Even assuming that's true, there's still a lot of stuff going on in the world that never makes it on the internet and never gets recorded by sensors."

Repeat of sentient phone network by Ray Bradbury ... back when it was just transistors and relays.
Ha’át’íísh baa naniná?
Azee’ Å,a’ish nanídį́į́h?
Táadoo ánít’iní.
What are you doing?
Are you taking any medications?
Don't do that.

Hydra009

#7
Quote from: SGOS on November 19, 2017, 12:49:48 AMI keep thinking about my first person shooters.  The first few times through a scenario, you get your ass shot to pieces.  But eventually, you know that when you get close to that tree a guy is going to pop out, and after that, some guy will start shooting at you from behind.  The enemy always does the same thing so you play as if you have some psychic ability to forecast events.  And even in games where the enemy is programmed to not always use the same tactics, its differing strategies are still limited to a few.

Challenging yes, but not really all that intelligent, but they call this AI.
Video game AI is really just an intentionally simplistic decision tree.  If player takes dialogue option #2, generate dialogue #2 then attack player.  If HP < 30%, use a healing potion.  It superficially resembles intelligence, but it's not really intelligence in the way most people mean it.  Extra Credits goes into this in more detail.

Primitive video game AI (weak AI) =/= artificial general intelligence (strong AI)

QuoteTry it this way; Artificial intelligence ≠ intelligence.  Just because they refer to each with the same word doesn't create equality.
Obviously, human intelligence and machine intelligence is not currently on anywhere even close to an equal footing, but that doesn't mean that artificial intelligence is a fictional concept.

AIs can currently beat the best human players in the world at Go, Chess, and Jeopardy.  They can perform surgery, pick stocks, understand human speech, drive cars, create music, etc.  And they're improving.

True, there is not yet a general-purpose strong AI.  But to say that there is currently no intelligence at all is simply untrue.

QuoteWe can have this conversation again 200 years from now.  Perhaps with different outcomes.
Perhaps.  But in such a scenario, this would no longer be an issue and the position that artificial intelligence is just a meaningless buzzword would be laughable.

Hydra009

Related: AI Effect (a method of discounting AI by declaring that whatever tasks an AI can perform do not require "real thinking" when they previously did when it was a purely human activity)

Atheon

AI beings would have to worship humans... their creators.
"Religion is regarded by the common people as true, by the wise as false, and by the rulers as useful." - Seneca

SGOS

#10
Quote from: Hydra009 on November 19, 2017, 12:52:42 PM
Related: AI Effect (a method of discounting AI by declaring that whatever tasks an AI can perform do not require "real thinking" when they previously did when it was a purely human activity)
I understand the debate between AI fans and the discounters, and I even admit I felt a bit guilty being a discounter, because discounting naysayers are always so stuffy being so uppity and all, but so be it.  Count me as one of the discounters. I may change my position, but that's along way off.  I see the debate founded on philosophical and semantic issues, which I think makes for a good debate, but also leads to both sides talking past each other.  I don't hold philosophy and semantics in high regard.

This does not mean I don't appreciate AI.  What they have accomplished so far with it is impressive and valuable. My issue is with the semantics.  Define AI as intelligence and the fans automatically win.  I can't bring myself to do that.  What computers do and what humans do are different processes given the same name.  Not to say the flawed intelligence of humans is better.  I would never suggest that.

Back in the 50s a friend of mine built a tic tac toe machine for the science fair that would always beat you or tie you, but as impressive as that was, even then I could understand the algorithm that it used and could somewhat imagine the circuitry involved.  It wasn't a scientific mystery.  Computers do it way better, but it still happens because some geek devised an algorithm and had an understanding of what he was doing and what makes it happen (I think).  There are indeed some serendipitous and accidental outcomes in programming in the category of bugs that turn out to be useful, but when the coding is studied, they can be understood.

Human intelligence, even dog, cat, and dolphin intelligence is a bit different than AI in that no one understands it.  Neruo scientists can trace brain pathways and locate thoughts in a persons head, but how those neural pathways end up being perceived as a thought or a memory of a visual experience, no one knows how the end product is derived from the scramble of electro chemical reactions.  And those that claim they know (usually theists) may have an answer, but so far are unable to support that answer with anything more than an assertion.

OK we may understand it some day, just as someday an AI will do the same thing.  But the jury is still out.  What we are debating here is a thing neither side understands, and lack of understanding doesn't make for good points in a debate.  To me the debate can only be decided when everyone agrees that human intelligence is the same thing as AI by definition, but as of now we have no way of knowing if that will ever happen or ever be understood.  That's why I see this a semantic issue, and why I won't concede to an arbitrary definition.

One more thing in regards to the link:  It claims that humans want to hold themselves apart from AI to reserve a special place for themselves, and I think many will do that.  I don't.  I don't hold human intelligence as anything more than an evolutionary artifact fraught with something like programming bugs, actual artifacts that make human intelligence so far from perfect that we ought to be ashamed of our species, its power to reason, and the pervading ignorance associated with it.  There is nothing special in that that deserves to be held on a higher level.

Is AI the same has human intelligence then?  Maybe, but no one really knows.  Until someone actually does, it's like holding a formal debate in an hurricane of confusion.


SGOS

Quote from: Hydra009 on November 19, 2017, 11:08:21 AM
Obviously, human intelligence and machine intelligence is not currently on anywhere even close to an equal footing, but that doesn't mean that artificial intelligence is a fictional concept.
I hope I didn't say I thought AI is a fictional concept.  It's a very real concept.  It's is artificial intelligence, and it exists in reality, not just as a concept.  But I don't think that's the debate.  Do you?

Hydra009

Quote from: SGOS on November 19, 2017, 03:34:47 PM
I hope I didn't say I thought AI is a fictional concept.  It's a very real concept.  It's is artificial intelligence, and it exists in reality, not just as a concept.  But I don't think that's the debate.  Do you?
I'm not sure.  Sometimes we seem to be in complete agreement and sometimes not, yet my position has not changed.  Then what has changed?   :headscratch:

My position is that our current AI technology is still in its infancy and current AIs are not general-purpose (narrow AI) and are well below the human level.  But assuming any rate of improvement, AIs are likely to reach human-level intelligence in the future.

In your post, you describe AI behavior in a video game - basically, a NPC running a predefined script with no variation ("enemy always does the same thing") and conclude that the "intelligence" in artificial intelligence is a misnomer, that it's really just number crunching, not intelligence.  Therefore, AIs cannot match human intelligence because AIs do not really possess intelligence.

I'm trying to sum up your position as best I can.  I'm practically quoting you verbatim in places.  So if I completely misread you, I'm sorry.  I'm just trying to understand where you're coming from and I only have your posts and an imperfect reading of what you're saying to work with.

So if you'll sum up your position in a few sentences, that'd help me understand tremendously.

Hydra009

#13
Quote from: SGOS on November 19, 2017, 03:27:36 PMOK we may understand it some day, just as someday an AI will do the same thing.  But the jury is still out.  What we are debating here is a thing neither side understands, and lack of understanding doesn't make for good points in a debate.  To me the debate can only be decided when everyone agrees that human intelligence is the same thing as AI by definition, but as of now we have no way of knowing if that will ever happen or ever be understood.  That's why I see this a semantic issue, and why I won't concede to an arbitrary definition.
I don't understand this demand.  To me, it's fairly obvious that human and machine intelligence is different in much the same way that human and various non-human animals are different.  We're not wired the same way.  We don't have the same capabilities.  But two things don't have to be identical to be comparable.

If a particularly smart elephant gets into Harvard, gets a degree in criminal justice, passes the bar exam, and gets a job as an attorney, I'm liable to consider the elephant to have a human-level intellect despite its brain not being the same as a human brain.

Machines are starting to do a lot of intellectual tasks that were previously solely the domain of humans.  Rendering a medical diagnosis, for example.  It's on this basis that I say they're progressing towards human-level intelligence.

Hakurei Reimu

Part of the reason that AI accomplishments tend to be discounted as real intelligence (the AI effect) is because the AI doesn't accomplish the given task in any way resembling the way a human would accomplish the same task. Deep Blue didn't beat Kasparov the way a human did. Deep Blue didn't calculate in terms of strategies and goals and looking a hundred moves ahead in a likely cone of responses like a human player would. Deep Blue looked typically somewhere around ten moves ahead, achieving twenty moves ahead in some cases. It never "thought" about the endgame except when it encountered it on its normal brute force search. It also thought in terms of scoring based on the value of pieces in particular positions. This is a completely foreign way of thinking for chess players in general. A chess game played by humans is as much White's creation as it is Black's, and this is not what happens in a match against Deep Blue â€" the computer's move is very much a mere response to the human's, however sophisticated that response may be.

The term "artificial intelligence" conjures up images of a robot speaking and moving in a manner indistinguishable from a human. We would be making a human intelligence, just by artificial means, and in a way, I tend to think that AI is part of the search for what makes humans intelligent, and this does not seem to be what AI research is producing. This is, on the whole, what puts people off calling AI "intelligent."
Warning: Don't Tease The Miko!
(she bites!)
Spinny Miko Avatar shamelessly ripped off from Iosys' Neko Miko Reimu