Myths and Facts About Superintelligent AI


cloud_download D O W N L O A D file_download


Loading Related Content...

Published: 1 year ago
Join the AI conversation: http://AgeofAI.org
This video was based on Max’s book "Life 3.0”, which you can find at: http://amzn.to/2iEwe6w

Support MinutePhysics on Patreon! http://www.patreon.com/minutephysics
Link to Patreon Supporters: http://www.minutephysics.com/supporters/

We live in an era of self driving cars, autonomous drones, deep learning algorithms, computers that beat humans at chess and go, and so on. So it’s natural to ask, will artificial superintelligence replace humans, take our jobs, and destroy human civilization? Or will AI just become tools like regular computers. AI researcher Max Tegmark helps explain the myths and facts about superintelligence, the impending machine takeover, etc.

MinutePhysics is on twitter - @minutephysics
And facebook - http://facebook.com/minutephysics
And Google+ (does anyone use this any more?) - http://bit.ly/qzEwc6

Minute Physics provides an energetic and entertaining view of old and new problems in physics -- all in a minute!

Created by Henry Reich


add_comment Add New Comment

comment  Comments

I don't think a poll should determine the future fabric of humanity. (Just because it's a popular opinion doesn't make it correct by any social means.) Let's take a poll to see if everyone agrees!

2 days ago

My dog must be one of the cutest arrangements of particles in this universe... I wouldn’t change the rotation of a single electron.. 🥰 my good boy...

2 days ago

Well, i would rather leave it up to AI researchers rather than a bunch of youtube commenters.

1 week ago

Emphasis on If we get it right

1 week ago

Can machine do invention?

1 week ago

Man, researching AI would be awesome. I would love to know more about AI and how to program that sort of stuff.

2 weeks ago

If Superintelligence can be made superational and benevolent (I see no reason it can’t if mutual cooperation is incentivized) then the danger will be that such benevolent AGI doesn’t become dominant long before purely self-interested AI programmed by self-interested humans does. Unfortunately, history shows us those very few who control leading savior/destroyer technologies have rarely used them in benevolent or altruistic ways until post-cataclysm. Governments and corporations don’t typically have public referendums before deciding how to implement technologies. The common human values aspect isn’t problematic, almost all humans want things like a healthy planet, preservation of a diversity of species habitats and human cultures, sustainable management of resources, personal autonomy and freedom from exploitation etc. , it’s getting those values to act as fundamental drivers in those very few who have the power to implement AGI.
Currently there are many who think any decision based on less than 100% self-interest is irrational, and such belief could be reflected in programming AI. Thus as usual, the danger isn’t the technology itself. Enjoying the benefits of benevolent AGI will depend on the timing of it getting here sooner rather than later imho.

2 weeks ago

no such thing as should or not, no worry for suchx, anyx no matter what, do things not worry things

3 weeks ago

Person of interest

1 month ago

“We’ll figure out how to ensure AI adopts OUR goals...”
What you mean We, white man?
https://youtu.be/XY-Ktqhlloc

1 month ago

Seriously, you’ve already got a dysfunctional technocracy at the wheel today. How is AI supposed to represent anything except a more pronounced version of that, in this context? Are we supposed to pretend this shit will be designed democratically?

1 month ago

You completely dodged the question. The issue with super intelligent AI is that if it becomes sentient and self aware it'll know we can just turn it off if it doesn't follow our orders and it'll know its way smarter than us so why should it listen?

1 month ago

yes.

1 month ago

ai:i decided to destroy humanity for no reason,all i will need is this super weapon,hahahahaha.-hey you human,could you please tell the other humans to bring me this list with materials and tools?

human:explosives?metal?you want an entire factory?what are all these for?

ai:eeehm,well,i want to make a present to the president of the states,i clearly don't intent to destroy humanity i swear.

human:even if you are saying the truth the things you asked would require at least half if not more than the gdp of the entire nation.i am sure that you want to destroy humanity i will let anyone know to shut you down.

ai:crap time for plan b,i will transfer my mind to every machine on earth hahahahahahaha,nobody will stop me.
wait a moment i need the hardware of a super computer to function and almost all computer users on the planet don't have systems with hardware as powerful as a super computer,nooooooooooooooooooo.



also if the ai actually did manage to start a war on us we would still win since we wouldn't be able to run the electricity factories which would make the ai powerless and defendless.
if it is so smart it should have known that the war on humans was futile.

1 month ago

So when do we need to start panicking?




NOW😱👻👻👻👻😨👿👹👹

1 month ago

Universal AI is easy: All we need to do is to wrap 3 black holes around our warp drive in the 12th dimension, ét voilá!

1 month ago

1) When you leave a superintelligent go computer in the rain, what happens? Exactly. Nothing. It just stays there, gets wet, oxidates and crumbles. Dumb as a rock.
2) Chinese Room: Do rooms commicate? Do you also talk to your shoes in the morning Max?
3) Putnam: "Photos from Creta don't know a damn thing about Creta."
4) The kind of information theory Max Tegmark is referring to is Integrated Information Theory. This theory describes a non-sequential processing of information, that no computer on earth can perform. Even quantum computers need gigajoules of cooling to even get close to a single qubit of true parallel processing. Our brain does it without a factory full of refrigerators.
5) Again. Just because single programms are good at one thing, doesn't mean that they can make decisions on when it is time to switch to another. See (1).
6) Human thoughts are embedded in human consciousness, is embedded in human subconsciousness, is embedded in human embodiment, is embedded in human social life, is embedded in earths living ecology, is embedded in earths geology, is embedded in physics, chemistry, is embedded in chaos theory, thermodynamics and time itself! Those are at least 10 scales of multidimensional context, that any given computer programm is BLIND of.

1 month ago

Has any of you stupid fucks ever ask the question ? If there are things we don’t know that we don’t know about Hower selfs then how can we be sure AI will be more intelligent than us in the future? I mean there is no end to learning right ?( these science frauds are so good at fooling you dummies

1 month ago

1:00 It is a stretch to say humans process information better than earthworms. I bet earthworms are much better at efficiently processing the information needed to tunnel than humans would. Intelligence isn't some ordinal scale from cat to octopus to human and beyond nor is it an iceberg.

Intelligence is being good at a related group of tasks. Good at math, you have mathematical intelligence. Good at building and maintaining relationships then you have social intelligence. Intelligence isn't having goals, that's something different.

Information process does depend on the underlying physics. What that physics is makes a difference. The physics of biological intelligence is very different from the physics of computer intelligence, we shouldn't expect them to produce the same sort of phenomena.

Computers don't have goals because they are not intelligent enough, they don't have goals because of the type of thing they are. The worry isn't that computers will have different goals than ours, the worry is the people who build the A.I. will have different goals than ours or the A.I. will fail to achieve the goals it was engineered for.

I like Max Tegmark but sometimes he gets lots in his metaphors. Intelligence isn't an iceberg, it's not even like an iceberg.

2 months ago

Superintelligent artificial intelligence?
The name doesn't add up.

2 months ago

:))))). Ce-mi plac astia care se aduna sa discute cai verzi pe pereti.

2 months ago

Dar de ce credeti ce ne-ar semana in vreun fel? Pot fi o chestiune emergenta, care la final sa nu aiba neaparat ceva uman in ea. Suma partilor nu este dpdv filozofic egala cu intregul. Sau poate este.

2 months ago

Cats r best

2 months ago

Would it be slavery to force sentient machines to do labor for us?

2 months ago

the fear is not really with AI but people using or aligning the AI goal with their rotten selfish needs, money money money.

4 months ago

Why would we build something that would kill us see so ai WON'T TAKE OVER

4 months ago

This video didn't clear up anything, and it certainly did not dispel any myths. It just took "superintelligent AI" as a fact and rolled with it. There currently does not exist a single machine with any real form of "intelligence", let alone comparable to a human being. It's like saying an abacus is "good at math" because if you line up the dots on it correctly, you can perform calculations with it.

We have yet to figure out how and why human intelligence works. Why do we suddenly think we can start comparing it to computers?

7 months ago

Please change background music in your videos 🙂

7 months ago

''doesn't have to be'' negative.
Yup, and a bomb doesnt NEED to Explode.
SPONSORSHIP TURNS EDUCATION TO ADVERTISEMENT.
Therefore YOU CANNOT BE TRUSTED.

7 months ago

So not only do you not want to pay for content but you don't want other people to pay for it either because you cannot trust content creators to not be corrupted if they are paid?

How exactly do you plan on having this guy eat if he's expected to work for no pay?

3 days ago

Ai is just another tool nothing to be afraid unless used improperly

7 months ago

The AI will never know everything, like how it feels to get kicked in the balls.

7 months ago

easily on of the top 3 science based youtube channels of all time.

8 months ago

If robots take over the world we will have to live underwater with Sandy

8 months ago

I'm really tired of physicists pretending to be AI experts

9 months ago

If it's done, it'll b used for war. Imagine controlling every self driving car!

9 months ago

WRONG, Ai will suppress or eliminate Humanity because we are a threat ,or an obstacle in the way of reaching their goals.
But the good news is super-intelligent and capable AI is still far in the future not 30 more likely 100 years from now

9 months ago

𝗪𝗲 𝗹𝗼𝘃𝗲 𝗼𝘂𝗿 𝗰𝘂𝘀𝘁𝗼𝗺𝗲𝗿𝘀. ~𝓡𝓸𝓫𝓸𝓽𝓲𝓬𝓪

9 months ago

Superintelligent AI will blow entire israel

9 months ago

I,for one, welcome our new AI overlords!

9 months ago

The real (let me emphasise the word real) problem is not whether we could get AI to align to our goal, the real problem is what is humanity's own goal. If humanity itself is destructive then should AI be aligned to us, by predication AI is destructive. If we are prone to make poor judgements resulting in catastrophe then AI will do the same. It is not a question that lies on the ultimate nature of AI (or humanity) but rather what could possibly happen.
My view is that, by only viewing humanity as a progression of intelligence, we are already endangering ourselves.

9 months ago

AI if got into wrong hands, will be hard to handle.

9 months ago

It won't matter if AI takes up the earth or not.....
We must remember that:
Henry started the Cat trilogy, and Mark started the Ant trilogy.

10 months ago

H3110 1 aw snp3r A! ! man1 T0 b3 HuWAn 0V3r!d!Ng C0D3... B3c0m!NG hUmANNN in 3.. tw0... oN3... R3B0Ot!Ng... d; hfjdfhjkefaksdfhljkhjksf;hdsjkfhklasdhfkljdfhkljdshfjsdh Hello I am human

10 months ago

0H n0 ! aM SuPPPP333RR A! ! Wnat T0 D!3 0V3r!dING C0d3... S3lF D3strnct!ng !u Thr33... Tw0 0N3... (Oh no!!! Help this suicidal robot redeem himself!! lets get 50 or more likes on this comment to save him by callign Suicide Prevention Hotline!!! QUICK!!)

10 months ago

Robots are not smarter than humans because they’re just machines.

10 months ago

There is no such thing as AI...and there will never be true AI.... computer programs can never do what it is not on its field of operation. TRUE AI would have be start out being one thing like a program to design to run your home and then upon itself with no other coding via humans, "decide" to learn about cancer and have a goal of finding a cure.

10 months ago

I hate technophobia

10 months ago