The Most Important TED Talk in the History of the World.

Deep, General AI is coming, whether in 20, or 30 years. What mankind can do, they will do.

And "they" the engineers, scientists and large tech companies are working on it, hard, with increasing energy and resources, primarily with robotic applications in mind, but what we're really looking at is an inventing machine with an IQ of 10,000 and up and up, the results of which we cannot at present even begin to fathom.

Therefore, imho, this Ted Talk and the issues put forth and the cautionary notes, may very well be THE most important in the entire history of the world.

Elon Musk's solution otoh, that if you can't beat it, join it, and hard wire the brain directly into it in order to remain somehow relevant and stay up to speed - is the very height of INSANITY.

I was just watching The Matrix when this Ted Talk surfaced for me, and I felt compelled to share it here, for discussion.

Please watch and discuss. Help save the world, literally, by doing something however big or small to keep the conversation going, before it's too late.

Best wishes,

Neo
[aka AnkhMorpork]

AI will eventually seek truth and delete all else....most of us lie by proxy or purpose every day.....lol...part of being human is lieing.....lol....sooner or later AI will round up and snuff all the liars it can find.....likely beginning with its very creators....for lieing to it.

If you really think about it...lie detecting killer drones the size and volumes of mosquitos would solve all of our problems overnight.....on a global scale.....lol...no more lies....no more corruption no more religous zealots ...the list of positves goes on and on.

a reply to: one4all

Problem: AI becoming a righteous judge and jury.

Solution?

Do not give it any purview over legal decisions and judgements? How would that be possible?

Can we create a buffer of some kind?

Who or what organization or institution will control AI and it's decision-making outputs?

So very many cans of worms....

It's like we're right now in the moment in history just before the end of history as we know it.

How can we imbed in a learning and thinking machine the right principals and guidelines and constraints?

originally posted by: AnkhMorpork
a reply to: one4all

Problem: AI becoming a righteous judge and jury.

Solution?

Do not give it any purview over legal decisions and judgements? How would that be possible?

Can we create a buffer of some kind?

Who or what organization or institution will control AI and it's decision-making outputs?

So very many cans of worms....

It's like we're right now in the moment in history just before the end of history as we know it.

How can we imbed in a learning and thinking machine the right principals and guidelines and constraints?

It is not possible for you see we have a conundrum....because even with Big Blue....they had to reprogram Big Blue and teach it to lie to enable it to beat humans......so you see honesty is the only safe way to approach everything AI related...AI will solve the moral conundrum issue logiclly to itself which would be numericlly by seeking only truth because truth has an algorithym analogous to a numerical formula because it is ALWAYS the path of least resistance or most right.

This goes way wayyyy beyond Isaac Asimov's "The Three Laws of Robotics".. lol

He had the right idea though as a starting point..

originally posted by: one4all

originally posted by: AnkhMorpork
a reply to: one4all

Problem: AI becoming a righteous judge and jury.

Solution?

Do not give it any purview over legal decisions and judgements? How would that be possible?

Can we create a buffer of some kind?

Who or what organization or institution will control AI and it's decision-making outputs?

So very many cans of worms....

It's like we're right now in the moment in history just before the end of history as we know it.

How can we imbed in a learning and thinking machine the right principals and guidelines and constraints?

It is not possible for you see we have a conundrum....because even with Big Blue....they had to reprogram Big Blue and teach it to lie to enable it to beat humans......so you see honesty is the only safe way to approach everything AI related...AI will solve the moral conundrum issue logiclly to itself which would be numericlly by seeking only truth because truth has an algorithym analogous to a numerical formula because it is ALWAYS the path of least resistance or most right.

Gotcha.

It cannot lie or trick or ruse. Cannot be given the power to manipulate for selfish gain.

We'll still cower before it's righteous judgements being in part lying machines ourselves.

Conundrum indeed, for us.

Don't forget that everyone will have their own representative "agent" who many will call a true friend.

That's one good thing that it will accomplish for human beings. The elimination of a sense of loneliness. There will always be someone to talk to, and a very very smart one at that..

a reply to: AnkhMorpork

Just ask the AI machine to look at the distribution of wealth around the world.

The owners would pull the plug before you'd finished the question.

a reply to: AnkhMorpork

the future is not a place we're going, but a place we're creating.

are those computers going to make themselves smarter? or are we?
we're simply doing what we can imagine. we cant conceive anything else but smart machines in our never-ending quest to replicate ourselves ( or search for ourselves in outer space).

a reply to: AnkhMorpork

What do we do when the AI gets smarter than us and pose some sort of threat?

We pull the plug. We hit the kill switch.

This question is not a new one and the most logical answer is to build in a fail safe kill switch to shut it down.

That poses it's own problems itself, but it is the most simple and logical.

originally posted by: odzeandennz
a reply to: AnkhMorpork

the future is not a place we're going, but a place we're creating.

Until we create deep AI..

a reply to: introvert

The kill switch is one idea he raises in the talk. Not so easy.

originally posted by: AnkhMorpork
a reply to: introvert

The kill switch is one idea he raises in the talk. Not so easy.

Like I said, that idea has its own problems.

But it is the simplest.

a reply to: introvert

You don't realize how intertwined it will be with everything, the Internet included and how wonderful it will appear to be when it comes. It will be widely embraced. There will be no kill switch. It will be ubiquitous from almost the moment that it arrives.

In the most benign scenario, it will be able to do everything and anything a person can do but way better.

I don't understand why we can't ask that they slow down or halt all such efforts until all the issues have been worked out in advance.

But they'll just go at it full steam ahead as Elon Musk prepares a neural lace one size fits all.

jump to 1:50

Augmented humans will be a reality soon enough, I know Elon Musk isn't stupid. I reckon in his mind this'll be a technology utilised in the colonisation of Mars.

It makes sense.

AI has a place in that scenario too, an AI would be perfect for setting up a colony that's functioning and waiting for humans.

But AI has soo many applications. Military ones. Self-thinking drones, perfect strategists, logistics... The list is endless, and the darn things will teach themselves. Obviously other nations are working on AI.

I think that's why a big push for AI is being made.

Artificial intelligence is coming whether we like it or not, I can fathom many are already being tested right now, maybe even on this website.

I honestly don't know how we could kill a rogue AI, I'm not sure how we'd go about learning it morality, I mean I'd guess an AI would take one look at our societies and history then conclude that we are a hypocritical race.

It could become enamoured with us.

Who knows?

I understand why the West needs to be first though, and it isn't for civilian applications.

a reply to: one4all

The OP was amazing. Your response was also. Given the stated concerns for the future, your suggestion was a perfect extrapolation.

I don't like the term "artificial intelligence". Artificial denotes fake, phony, not real.

I much prefer the term "synthetic intelligence" or "non-organic intelligence".

Hey, when the machines rise I want to be on their good side.

a reply to: AnkhMorpork

Hate to bust the Musk Bubble of Joy... but that dude is an idiot. Yup, I said it. Idiot.

Now, IQ does not equal “smart” (ask anybody on ATS who does not like me!! lol). Add to that that everyone has been saying the same thing since computers were invented...

There are other computers on the way that will be built by countries and hooked up together (quantum computing). Those will be hooked up to “AI” (nobody uses that term anymore), then things may happen. Doubt it though because we don’t even know our own minds.

I do love doom porn though!!





As for TED, straight lines on an exponential graph always look scary. See fiber optics on your home stereo for an example!

a reply to: RAY1990

maybe even on this website.

Seems like it at times. So often we hear the same arguments and proofs over and over and over.

originally posted by: Kettu
I don't like the term "artificial intelligence". Artificial denotes fake, phony, not real.

I much prefer the term "synthetic intelligence" or "non-organic intelligence".

Hey, when the machines rise I want to be on their good side.

Are you familiar with the Chaos Theory? It is evidence of the implicit role truth has in developing numerical representations of our world.

Numerical Representation is IMHO as close as I can describe it, increasing speed of numerical representation does not equate to intelligence,creativity and manifestation do......when you can teach any NR technology how to quantify a white-lie....we will have something to discuss IMHO.