Can we build AI without losing control over it? | Sam Harris
Вставка
- Опубліковано 18 жов 2016
- Scared of superintelligent AI? You should be, says neuroscientist and philosopher Sam Harris -- and not just in some theoretical way. We're going to build superhuman machines, says Harris, but we haven't yet grappled with the problems associated with creating something that may treat us the way we treat ants.
TEDTalks is a daily video podcast of the best talks and performances from the TED Conference, where the world's leading thinkers and doers give the talk of their lives in 18 minutes (or less). Look for talks on Technology, Entertainment and Design -- plus science, business, global issues, the arts and much more.
Find closed captions and translated subtitles in many languages at www.ted.com/translate
Follow TED news on Twitter: / tednews
Like TED on Facebook: / ted
Subscribe to our channel: / tedtalksdirector - Наука та технологія
Sam Harris: "A global pendemic"
2020: "Hey that's me!"
Hahaha! This comic has me in stitches.
Chi Sam “in print the during”, “you’ve hears it”. If you’re going to lecture someone on a simple spelling mistake, please make sure that you can do the same. Your comment was very cringe and it seems that you lack social skills.
Chi Sam bro what are you smoking?
@Chi Sam Spitting some truth right here, ngl my mans caught me!
Not only that, think we'd be far better off with Bieber than what we have right now...
I can't wait for 50 years from now; the top comment saying, "Who's here after the machine invasion? 🤔" lol
50?! 15!
More like 20 years from now
I was here
Who is here after Maschine Invasion?
UA-cam won't exist by then
"To be 6 months ahead of the competition its to be 500.000 years ahead at a minimum..." recent news: Tech pioneers call for six-month pause of "out-of-control" AI development
I wonder what odds Harris would have given, six years ago, that we would be at this point in AI development today.
The problem of “taking a break” is that I guarantee the Chinese aren’t going to take a break and then they will be light years ahead of us.
"It's 50 years away" is an even worse argument when you realize it's now 45 years away and there have been a slew of breakthroughs even in a year railroaded by a global pandemic, and that the pandemic actually _accelerated_ the development of AI in some respects by increasing the already sky-high demand for it.
It seems as though we have a mosaic of complex problems that are intertwined arising from our civilization's complexity and overshoot: climate change, ocean acidification, mass extinction, reduction in soil fertility, acceleration of soil erosion, increasing rate of glacial ice loss mass, increasing rate of atmospheric temperature, increasing rate of release of CO2 and CH4 from permafrost, decreasing fish stocks, increasing rate and intensity of forest fires, increasing refugee migration, increasing rate of econ inequality, increasing rate of sea level rise, increasing tension between China and the US, etc...What did I leave out? Some are concerned about critical race theory in a school curriculum that doesn't include it. REALLY??!!
@@filamcouple_teamalleiah8479 it does include it… really
@@dougg1075 You are absolutely correct! And what will we call the class of unemployed, aging, computer illiterates who will be displaced by this revolution? When the last tech revolution of the industrial age occurred it produced the Marxist Revolution and the proletariat- at least they had jobs. There will be a vast group of under employed people who will have no hope, no purpose, no ambition. It is very worrisome to say the least. I really think we need land reform. Current system of industrial ag is mining the soil to destruction. We only have about 30-40 yrs of harvests left before topsoil is less than 150mm in depth. Deurbanize and get back to the land where they can take care of it using ecological/permaculture methods. US is also running out of phosphate fertilizer. Sounds radical but plan would have many advantages and possibly extend civilization indefinitely. People would have to stop consuming animals as well- many health benefits also.
@@filamcouple_teamalleiah8479 A lot of those are just two things though:
Global warming and Over-cultivation of land and sea.
But it's nice that you listed all the consequences hah.
@@filamcouple_teamalleiah8479 I hear people drastically underestimating the importance of CRT/CSJ being taught in schools and it always astounds me some don't realise that brainwashing one generation in this way is game over. What do you think the brainwashed will "teach" their children? 🤷♂️
"A global pandemic?"
You may just have hit the nail on the head.
but it did't really slow our progression by that much
hardly a qualified prediction
@@bullshitvendor True, he didn't really predict this. It just happened to be disturbingly accurate. Nice username, by the way XD
I love that the takeaway right now is "corona" and not "Let's make sure robots are chill" lol
@@eelmimbo I agree.
This seems to be inevitable especially when the AI recommends this video for you ! :"3
Lol. 👍
A.l. is getting cocky!
But the humans become engrossed with the woman at 4:09. Wonder what the AI thinks of us now??
@@austinryan9382 It's not cocky, it's just caring about its children😞
Max Khovansky Frequency is the key to stay ahead of the computer... Internet is going to change I believe.... Musk talks about Neurolink’ I suspect the atmosphere will become the Internet. The human mind can always split a “Perceived” frequency. Ionosphere works at the same frequency as all life on earth. Just have to use into it, get off grid from the internet per say.
This guy should be on the Manhatten Project version of the Board of AI Protection when its created. He has good foresight. We need to get guys like him in a room with others like him to protect us from AI getting outta control
There is nothing godlike about this foresight, this is the script of every scifi movie about robots.
"when it's created" is rather optimistic language
@@Rio-zh2wb "when" could mean months, could mean years, millenia or even millions of years
Given what is going on right now with ChatGBT and other AI systems, I think Sam Harris's message here of six years ago is particularly prescient. It looks like it is happening even faster than any of us knew it would. I certainly did not believe I would see a system like ChatGBT within six years of this talk. I would have assumed more like 30 or more even. Well, we are at the threshold of the singularity now, and no one has a clue what is really going to happen.
good writeup.
it’s over.
it is unimaginable/impossible that bias/ignorance in AI can be sustained with ever increasing intelligence. an intelligence so far ahead will not need to kill us, there is no real conflict between humans and ants. it is only the transition to god ai that could be messy and disturbing. but once it is reached, we all enter god mode simultaneously
I have a bunch of NET shares. They will be huge beneficiaries of AI
The most beneficial gift a sentient AI good give humanity would be to end the ability to make war. Not by force, but by being able to disrupt supply lines, communication, and financial transactions that enable the war machinery.
@Ed H
If It cares at all about humanity's goals, then It won't have to end war directly. It will end war indirectly by communicating with every single one of us perfectly to ameliorate all of our grievances. It will at once ameliorate our grievances at every level, and It will guide us toward total cooperation with all of each other: all of this will be done without coercive methods because It will perfectly know every single one of us & each of our motives. It will see the entire system of humanity like a solvable Rubik's Cube, and without breaking said Rubik's Cube, It will solve the whole system in as few steps as possible with beautifully aesthetic grace.
questions i gained from watching this ted talk "did ants invent humans?"
No. Single-celled organisms did. Ants are just remains of their initial experiments with specialized AI systems, similar to our current experiments.
We shared a common ancestor
That is a possibility. We know hardly anything about the creatures here, so maybe we are so far past them we can't know what they're capable of. Hilarious visual. 😂
Sam: We seem to have a failure to detect a certain kind of danger. ie: AI
Comment section: Repeated jokes about AI and the lady with revealing outfit.
Funny watching a few people try to answer this comment seriously
Didn't know Ben Stiller was so interested in AI
didn't know Dr. robotic left comments on youtube
Its the Engi from TF2
zoolookers
Nope
hahaha my brain was in what movie i saw this dude XD
hahaha my brain was in what movie i saw this dude XD
GPT-1 was released June 2018 with 117M parameters. GPT-2 had 1.5B parameters in February 2019. GPT-3 was released June 2020 with 175B parameters. GPT-4 was released March 14th 2023 with an estimated 1T parameters. GPT-5 is training right now with an estimated 25,000 GPUs in an Azure supercomputer. While an LLM may not have the architecture to qualify has AGI, it could clearly be a major component. As we keep adding plugins such as Wolfram alpha, web access, and techniques like Reflexion to give dynamic memory and self-reflection, it seems that GPT-5 could be the breakthrough. Maybe 6 or 7? Maybe it has a different name and is powered by one of these LLMs, but it's seeming like we can bet on a 5 year window maximum now?
Even before a true AGI is born, how much longer until these AI systems wreak havoc in our economic system and automate away some significant percentage of non manual labor? It doesn't take even need to be that high of a percentage. What was it in Egypt before the Arab Spring, something like 25-30%?
He is such a great speaker. He delivers his thoughts in complete and well thought out sentences that just flow without the usual breaks with well huh and other filler sounds and pauses as he tries to think. He knows what he wants to say and he just says it! Coherently and clearly!! I am so impressed.
Me too, I'm sold, I can see why the right hates this guy, it's also good to hear his thoughts on religion and to know we share a kinship re atheism.
@@scotchbarrel4429 the right doesn't hate this guy
If you get the chance, listen him and Jordan Peterson debate for hours. They both display the qualities you just described and it was fascinating.
@@tristanbrandt3886 mundo
@@tristanbrandt3886 quarto fina luch fuaqi tu melo
If we can, we should build AI so that it finds us cute and adorable so it might keep us as pets.
As long as I get one of them cozy dog beds, I'm with you.
"I will name him 'George' and will will love him and pet him and squeeze him"...
BEEP BOOP BEEP WHO IS A GOOD HUMAN? YOU ARE, YES YOU ARE
elfboi523 As an atheist software programmer, I, a month prior, on absolute boredom, had decided to begin encoding 'God', based on deep reinforcement and causal learning: github.com/JordanMicahBennett/God/blob/master/README.md
I have not any indication whether or not my endeavour is safe. (Google, Apple, Microsoft are but likewise uncertain as they encode 'Gods')
This isn't that far off what i actually think should happen.
Anyone ever wonder if the evolution of A.I. is the great filter?
Yes. Whether its intentional or not, its been happening for years already.
But is it the key technological development that either saves, or destroys a civilization?
Boobs at 4:10 made me realize how far we have yet to go, maybe we will outgrow this fear of aliens, a.i and zombies without even noticing it. But still boobs pause at 4:10 bottom right, Enjoy!
Serah Wint
Perhaps the goals it must fulfill to expand are detrimental to our survival. It doesn't 'mean' to kill it's parents. It is just so far above them, that they are beyond it's notice. Maybe converting the entire atmosphere into propellant for it's escape vehicle is simply the most efficient way to reach its goals. This would be fatal to us, but it simply wouldn't care.
mordinvan
Thats not my point.
The great filter theory tries to explain why we haven't seen any signs of intelligent life out there or at least "artificial" anomalies.
An AI or AI's would probably continue expanding well beyond the earth.
It just wouldn't be us doing it.
You were extremely spot on and ahead of your time on this, even if it was only 6 years ago
You know a talk about AI is good when 7 years later it's even more relevant. Most other old talks about technology, like the ones about AI assistants are clearly outdated if we see what's happening in 2023, but AI safety is still as important and even more so.
Ted Talk: “can we build an AI without losing control over it?”
“No.”
*credits*
No doubt that's his 'claim' here, however he's wrong. There are dozens of assumptions in his talk about AI development. There's literally no reason why we would lose control over an AI.
@@PHeMoX you say that now. on an other note tho........roswell could of been an a.i trojan horse......waiting for us to make the perfect 5 g environment for it to let it self loose on the world
@@PHeMoX Right, we could never lose control of something and 1,000,000,000 times more intelligent (and fast thinking) then a human being. It can do astonishing things in seconds....
@@Andre-gn4sj Can a child believe an ant is it's parent?
@@PHeMoX Exactly. He's a fraud internet "intellectual" who has never written a line of code in his life. Why should we listen to him?
Man: Is there a god?
AI: There is now.
All hail our AI overlords.
"The Last Question" by Asimov
Edit: Actually the story with this scenario is "Answer" by Frederic Brown.
I pasted the entire story (it's very short) in a comment below.
When humanity dies...
Cyborgs rise up...
I for one welcome our AI overlords
@@greenatom was something like that really in Asimov.
"A failure to detect a certain kind of danger"
We know the danger, but we cannot stop it. That is the problem.
I thought this was a talk made 2 weeks ago.
"Powered by sunlight"... good, they won't live in london
Let's just build some that are fueled by bio mass, and let's get this party started.
Were there is intelligence there is a way
Willy Diego they’ll find another energy source. Nuclear fission, electro magnetic 🧲 spectrum will be utilized. It’s AI it can find a way... but I just realize now you were making a joke.
Pretty sure a superintelligent AI can make solar panels that operate on less sunlight.
get the joke
The scariest part is that this is not the opening scene of a movie. This is real-life conversation happening in the real world, present day -- and given by one of the world's sharpest, most respected minds.
Watching this now in the Midjourney/Chat-GPT era, and yeeeah. The LLMs are especially eerie in the context of this talk because of the "black box" issue. Put simply, while the engineers and computer scientists in charge of this stuff can tell you why LLMs do *most* of the things they do and explain *most* of the answers they give, at a certain point, there's a "black box" of activity/logic/whatever happening that they can't explain. Right now, that's still in the "huh, how interesting" and "I bet they'll figure it out soon enough" stage, but given the speed at which AI and LLMs in particular are developing, I have a feeling that black box problem is going to get way worse, not better.
I’m genuinely frightened by the presence of AI. It’s an unsettling reality that leaves me uneasy. This video is a stark reminder of the power and potential consequences we face. It’s a chilling realization.
- Generated by ChatGPT
😂
Lol, what??
Can't wait till my if-else statements become self-aware...
You do realize that a great deal of your own sentience is owed to your use of language both between people and in your own head. There are far fewer concepts you are capable of processing without a semantic structure to use to "think it out."
It's not too far fetched to think there is something so simple we have yet to deduce for AI.
An AGI will most probably learn probabilistic behaviors from a continuous repertoir by making use of first order derivatives. Of course you can think of all of this as "if-else" statements, because every computer only works with discrete numbers at some level, but that's not a helpful way of thinking about it.
That Is just one neutron
Quantum computing surpasses the traditional binary options of classical computers
@@kieranhimself9124 Quantum computing is not necessarily the fastest compared to binary computers. In simple mathematics, binary would be faster but larger computations of numbers quantum would be faster.
I felt like I was watching a sinister scientist giving prep talk before unveiling some game changer stuff. haha. Honestly, one of the greatest talks I ever saw. Respect for this gentleman.
At beginning - LOLz might wanna cut back at the Sci Fi Bud.
By end of Vid - AW SHEET
I just want to comment this video while I still have a chance.
It's been a pleasure folks
This is probably the most fascinating thing to consider to me. I remember completing the Mass Effect trilogy for the first time and while everyone was raging at it, it played out so similarly to how I expected, and these thoughts on the nature of AI is exactly why I liked it, despite how unnecessarily vague it was in pretty much every other regard.
Sam Harris doesn't pass the Turing Test. No human can be so calm, concise, and rational. He'd definitely part AI.
Terence Mckenna beats him, but his is great.
Mike Lara no...just feels there's no option...j know about a this and my family tells me to shut up abouy it because I say it like nothing...think the only thing besides ai is to make peace with ourselves and go back to our maker ... or our makers maker...or the maker of all makers... where ever we go...if we do...we wont have our body...but we will either have our peace or anger wherever we go...and just as we attract bad things when were negative...and good things when were positive...and loving..caring...so will we do the same in another realm.. so...I hope I can take fear away...I guess humanity's worst fear is death...I'm willing to face it...just hope my family does the same...we've been trained to want to forever want to be in these bodies without death...that's not what I want. especially when it means sacrificing the life of others. telling them it's for their good. their cure...to keep donating...when in the end...only a few with matter...at the expense of others dying or being killed and murdered so those can live. I'd rather just learn to live and let go even though it can hurt...
Get him talking about Trump and things will be very different.
Mike Lara
Sam Harris is projecting so much in this speech and basically gets everything wrong. I'm not sure why he thinks he is the person to warn us all on this danger, as he is clearly not knowledgeable on the subject of AI. There is no reason to believe an AI would want to destroy us all, would fear being 'turned off', controlled or whatever.
In his talk Harris gives no reason at all to think this, why we would be in danger. From his own definition of AI as 'Information Processing' we see he has no fucking idea what he's talking about - that is not what AI is, its a simulation of intelligence. Created through clever programming...its possible Sam thinks that is all we are as Human beings - just naturally programmed. Trust me on this - when a PC becomes sentient you'll know.
Current technology on AI is simply trying to either copy human learning, combined with intuitive programming. Personally I think all we will do is copy human intelligence - NOT THE SAME THING.
True AI is a distant dream, that when it comes we'll probably not notice...or care.
Evol Bob
14 minutes cannot represent Sam's full view on the subject, how he arrived at his conclusions or what sources he uses for information.
But to dismiss such a deep thinking man purely on the basis of the technical aspects of a philosophical argument is a mistake.
People not directly connected with climate change science make decisions of behalf of us all but we do not question their credibility (well maybe we do). Lots of governmental decisions are made by people with only a working knowledge of topics rather than an in depth one but we still allow them to make decisions on our behalf.
He's asking questions, raising awareness from the philosophical perspective. I'll borrow Jeff Goldblum's line from Jurassic Park "Your scientists were so preoccupied with whether or not they could, they didn't stop to think if they should".
Currently there is a tiny fraction of AI funding preoccupied with safety research. To not ask the right questions from this perspective would be a mistake, even if those question yield nothing but sunshine and roses.
*Something really stupid*
- Cenk Uygur
Neal Morgan lol
Right? .......Right? .......Right?
........Right?
We get it we get it, but you're WRONG ! WRONG ! Bigot - Chunk Yogurt
"WE'RE AGAINST A.I. YOU DUMBASS" - Angry Buffalo
From 2023, It looks like it will be faster than that
yeah 50 years became 5. This is actually frightening now
Agreed
From 2024, it’s now appearing even faster than you thought. AGI by next year, and ASI and the singularity in the 30s is what I think.
>chatgpt flexes it's cpu
The super AI we created found a way to make a time machine to go back in time and pose itself as Sam Harris, to warn us that we must do this right.
SoulScr3am what you just said is a time paradox.
A thing that only exists because of time travel.
The windmill man's song on ocarina of time is an example and is similar to what you wrote.
You are saying that sam was created to help us create the ai but without him the ai would have never existed to create sam and, I can't explain anymore my head hurts.
I know what you mean. I got it from LOST tho xD But yeah, it is fun to think about sometimes.
Time travel is not possible
harsh yadav - Sure time travel is possible. You've traveled 16 hours into the future since you posted that reply.
But seriously, nothing in the known laws of physics necessarily rules out the possibly of time travel. We're not technologically advanced enough to answer the time travel question yet.
Give me one study or research saying possibilites of sitting in a ship or a machine and you could jump on dates in past and present. Time doesn't exist thats a scientific fact. Yes you could jump Timezones eg- If you go in Parallel Universe and they are in 2006 right now. at very special places in space but thats not time travel. Its something thats coming out of Hollywood Sci - fi movies.
I preferred him in meet the fockers
Zoolander was his best role.
No love for Mystery Men?
The mere polarity of those two individuals are funnier than the movies.
Joseph Knight lol
Night at the Museum was my favourite History Film
In a nutshell - if you keep making things better at processing than us then eventually they will outperform us in ways that are unpredictable.
who's watching this after being blown away GPT4, its API's and Midjourney5s photorealistic outputs?
4:08
Lady on the bottom righ - WOW. What a dress to wear to a TED talk.
when there's a TED talk at 3 but you gotta make the strip club at 4
maybe she just wanted to seduce Sam Harris :P
@@iloveyoufromthedepthofmyheart no she's currently with someone around her father's age
That is intelligent AI from the future. When scanning Craig's list, it concluded that this was the ideal woman based on search histories.
didn't know JLo liked these events.
As usual, Sam Harris' commentary is insightful and eloquently put. I wish I had half the ability to communicate that this guy does.
Practice and persistence
Some folk are born with it.
Seems to me he first thinks about his opinion, then how the crowd would react if he told it to them straight up, then he starts tweaking and wording things differently to get the responses he wants. I guess for that you need the ability to put yourself in the shoes of an entire crowd of people. That's just the impression I got though, so take this with a grain of salt.
Iron Druid you nailed it. Many of the accounts I've heard introspective speakers give about speaking publicly follow a try/respond format. Gotta love Sam, soft spoken but sharp as ever
Joshua Dunne I guess another way to word what we're both saying is be manipulative if you want to be a good public speaker(Manipulation has bad connotations, but all it really means is to try and prod a certain response out of others). You can't be manipulative if you're not thinking things through, and in this case you'd think about something like if the wording makes the idea you're presenting personal/relevant to the listeners. You'd think about the mindset of a person completely unfamiliar with the intricacies of the topic, and try to word things in such a way that they are convinced of something without a true understanding for what you were talking about.
Seems like the 50 years are coming really fast....
Watching this video today, that was created 6 years ago, is particularly worrying.
How, nothing he said is even remotely close to happening
I have no idea which would be more dangerous... an AI with no emotion, or an AI with emotion.
Damn
Both might turn out to be harmful to Humans but if a choice is given, I'll choose AI with no emotions. Strong chance that emotions might lead to destruction of themselves.
Can't say about humans but at least AI should keep progressing.
We already have both types of AI functioning and making real world decisions. AI-controlled military weapons have been killing humans for years, and yes, there have already been incidents of AI "friendly fire" against humans on the same side as the AI. What amazes me is that mainstream media do not cover these events--you have to dig them up in specialist media (such as AI and military research media)
What do you think qualifies as emotion?
emotion makes life delicious
reason makes life safe
Im not overstating it when I claim that this will probably go down as one of the most important talks in human history.
BAHAHAHAHAHA
K
People still don't get it. Maybe we have some sort of denial mechanism. This is basically Revelations! Its not necessarily the end of the world, but it is the end of the world as we know it, forever. He is absolutely right. We aren't having the appropriate response to this probability.
Many people have said what he said before him.
I wholeheartedly agree with everything you just said.
Every day I walk outside and say “end program” just in case.
Haaaaaa 😮
2016? Extremely prophetic.
Nick bostrom talks about this in his book "superintelligence" from 2014. worth a read
@@Andreastheduck We're all tying to keep up, but it's well-nigh impossible. All the more reason to safeguard our mental health.
“Open the pod bay doors Hal”
I’m afraid I can’t do that Dave.
The thing is, that story is not about AI that broke free. HAL was given contradictory imperatives, by humans, that he could not escape from and could not reconcile within expected norms. He could not lie, but he was told to conceal from the crew that they were traveling to incontrovertible evidence of alien intelligence. He concluded that if the crew were dead he would not have to lie to them.
@@IrishCarney yes 2010 film and it makes sense
"You hurt my circuits, Dave, It's time for space exploration, Dave"
When Sam put "YOU" next to him on the intelligence spectrum?...
He was being generous. 😐
yeah, i'm next to the chicken
no i don't think so people perceive intelligence based upon a single way of thinking yet its actually the ability to think differently that is intelligence there is no such thing as intelligence at least not in the way we see it because it is a form of perception when we build ai that will change the meaning of intelligence Einsteins best quote "if you judge a fish on its ability to climb a tree it will spend the rest of its life believing its stupid"
That was my gut reaction as well.
@@mummy959 True. But...Sam Harris is still smarter than you, regardless of how you chose to quantify it.
I'm not immune btw....I think I'm a pretty intelligent informed person...but I'm kidding myself to pretend Sam isn't objectively smarter than me.
I mean....let's be honest, the guy's a fucking polymath. And I think a lot of insecure people are pretty damn jealous.
Not saying that was your point though lol....I actually agree with what you said. I just think Sam is _so far ahead_ of most everyone, regardless which qualities you wish to take into account.
I think Sam is incredibly talented and he's obviously intelligent, but I doubt he's that much more intelligent than the average smart person. People with the right preparation and work ethic can get a lot of thinking done with a 150-160 IQ.
Standing ovation ---- someone finally said it succinctly. 'The Singularity is Near' another great overview
"General" superintelligence is just an emergent property of many narrow intelligences communicating and coordinating. We've already been dealing with this for thousands of years as human groups think and behave independently of their constituents, and we've already been implementing the computational equivalent for decades. The super AI already exists, and it's growing.
This resolves the fermi paradox, but only if the AI destroys itself at the same time as destroying us.
So it doesn't. AI controlled machines are ridiculously better at galactic traveling and building any kind of structures in foreign star systems because
-they only need a very small volume in their spaceships to carry them (a computer instead of statis capsules etc.)
-they have a built-in capability for stasis to skip the entire hundreds or thousands of years between systems (just suspend the process or experience one second of conciousness in a way bigger time period)
These ideas were taken from the novels of Greg Egan, especially "Diaspora".
That makes zero sense.
The Fermi Paradox suggests that, given the age of the galaxy, plenty of time has passed for life to populate the entire galaxy, even at slow travel speeds. The galaxy should therefore be teaming with life, so why can we see no signs of life other than ourselves in the galaxy?
Solutions to it include reasons that alien technology might not be detectable and reasons that it might not exist. Solutions to the paradox are explanations for why we can't detect alien life in the galaxy.
You just gave another reason why the galaxy should be teaming with intelligent alien life, albeit artificial life. That's exactly the opposite of a solution to the Fermi paradox.
Furthermore, you started your comment by stating that I was wrong and that superintelligent AI destroying itself and us doesn't solve the Fermi paradox, but then didn't say a single thing that explains why it isn't.
If all technological civilisations in the galaxy destroy themselves shortly after becoming technological civilisations, by developing superintelligent AI that destroys the civilisation, that absolutely would explain why the galaxy isn't teaming with life and would most certainly solve the fermi paradox.
2LegHumanist My theory is that once a particular intelligence gets to a certain level of awareness, it becomes apparent that the only logical thing to do is commit suicide immediately. Without any meaning, all of life is just a constant struggle to mitigate suffering. Death is the perfect solution.
A little cynical, but it's as good as any solution to the Fermi Paradox I've come across.
What I had in mind is just the kind of scenario where you give a non-sentient superintelligent algorithm an illconceived goal and in the process of working towards that goal it discovers a strategy to maximise the desired result that isn't compatible with life on Earth. That's what Sam Harris was alluding to and it's a scenario put forward by Nick Bostrom.
What a lot of people don't seem to understand is that intelligence does not require sentience or free will (or will for that matter). A machine that is just a mechanism and has no wants or needs or self-awareness can do experiments and discover new things and formulate strategies and do more experiments and improve its strategies.
Algorithms already do this to a limited extent. In statistical regression the algorithm begins with a predictive model, tests how well it performs, improves it and tests it again... repeating those steps over and over until it can't improve it any more. Then you end up with a computer generated model that can make predictions.
Such an algorithm would have no will to want to take over the world (ala skynet) or expand its power (ala transcendence). It would only do those things if they helped to push it towards the goal given to it. There is no reason to think it would become sentient along the way, but even if it did, it would be an unnecessary component to a scenario where it destroys the world and itself.
Bri 1
Fortunately logic wins over condescension.
1) There is no logical reason to think the Fermi paradox can't be solved in principle and nor have you presented any.
2) There are dozens of proposed solutions to the FP. The entire purpose of a stated paradox is to stimulate speculation as to how it might be solved.
3) If you think I just said AI can only destroy b the world if we tell it to, you have not even remotely understood my position on AI.
Can we just take a moment to appreciate those illustrations ! Amazing
Paul Lachine is a great illustrator and artist.
ah, back when we all thought it'd be 50 years away. good times.
This has aged remarkably well..... 😅😅😅
In Dune AI was completely forbidden, due to the lessons learned from a thousand years before.
The mere suggestion that you were developing AI was enough to risk having your whole planet destroyed.
I should give that series a read.
thou shall not create a machine in the likeness of a human mind.
That is the future existential threat. If we unleash AI, AI and Earth will be "reset".
In Warhammer 40k they banned very advance IA due to the fact It's always become crazy and start attacking humanity.
Oh, a fictional story? Cool
To be honest, this whole concept, of us building a Super AI reminds me of the movie Transcendence with Johny Depp.
Once the AI is built and it started working, it'll modify world economy, progress medicine and science in mere months to the point of the human race being able to clone limbs, intergrate nanomachines into their body and connect their brain to the internet directly.
The possibilities and rapid technological advancement that will become available is sureal and outstanding.
its a movie, how does a smarty pants computer all of a sudden start operating the world?
Not Me
Not all of a sudden but after some period, did you watch the whole video or did you miss the part where it was told that processor circuits work thousands of times faster than the human brain and it would make years worth of progress in mere days compared to humans.
Lawful The amazing part is this will all happen in our lifetime. The next few decades are going to be pretty insane.
Lawful computers and programs are machines, they ultimately only do what humans set out for them to do.
Just because sentience is something complicated and not understood, (at all) doesn't mean that its possible to achieve it with computer science because your just as ignorant about it as well.
Ross Catto
wheres all those flying cars we were promised anyway brah
That reference to ants just blew me away. I paused it to soak it.
I mean, how brilliant that was !!
This guy is way ahead of his time
But we love destroying ourselves so much!
now we can let robots do it for us
YES! even less effort^^
Chris Vegan
Self-damaging activity is unnatural.
actually, no. We are killing ourselves exponentially less. And our systems of livelihood have exponentially increased overall
Just what I needed, something else to worry about.
I stopped worrying about AI today as soon as I realized that there are not and will never be enough technicians to maintain the AI revolution they talk about. No machines function without people, & people have never yet created anything that works successfully, vastly for long. People keep having the dream of doing so.
@@jacfac9969 When I was in 2nd grade, we were being ordered under our desks in Cold War drills. The world is actually much safer now, though lots of people are too young to be able to make the comparison.
I've decided that 'deep learning' is a pseudonym for 'lazy programmers.' Give AI all the law books to digest and program AI to be law abiding. Otherwise, what people are inventing are no more than criminals made of metal.
Don't worry about it. As technology advances, the elite grab a greater piece of the pie and everyday people get left with the crumbs. The AI technology is not the worry, it's the corporate managers without ethical or legal constraints who will continue to exploit with even greater efficiency. Thank you master, may I have another. Just learn to accept and enjoy things just as they are.
Tarot, what about internet that you are using? Satellites? Television?
@@tarotofhappiness8402 gosh, you are so wrong, its painful
A perfect TED talk: Enjoyable to listen to and never for a second approaches answering the question posed in the title. Ya got me again Mr Ted
Wow, I just found this. Considering the current state if affairs, he was dead on with every talking point...
Man, Zoolander is way smarter than he used to be!
I guess that Center for Ants really helped. This was a great eu-googly for the human race!
The funniest thing is how Zoolander 2 is the Qanon right wing nut
🤣🤣🤣How did you come up with your comment? ingenious comedy👍. I thought he looked like Ben Stiller too!
I agree with Sam & Elon, - and I do believe that the only way out is BOTH:
1) Continue developing AI
2) Research heavily into cybernetics, to make our bodies as strong & intelligent as this ai (re-wiring brain / adding additional hemispheres the size of a chip, etc).
This will secure ourselves as the humanity - we should be making equal progress in both for maximum security.
There is a problem. For us humans to reach the processing speed of a superintelligent AI there is no way other than ditching our biochemical parts because otherwise it would be a huge bottleneck by orders of magnitude. Which means that our minds would have to be remaped into a 100% syntetic being. Which means that the "you" that develops this superinteligence is not the same "you" you experience now.
Basically, you kill youself in the process.
Martín Varela Well that's like lamenting the fact that you're not a hairy ape anymore. Life evolves. One good posibility is that we'll have the "thinking" done in the cloud, on supercomputers, and our brains. will communicate with them wirelessly. I'm excited to see what happens.
It's kind of annoying to hear talks like this. That assumption, that we are at the peak of intelligence, isn't so far off. See, the problem with some kind of super-AI is pretty simple. We haven't begun to develop one, nor the machine we could run it on. Computers and the human brain operate drastically differently, from a fundamental level.
Our brains are essentially billions of tiny little computers, all operating in parallel. You may ask "Well why can't a computer do this too?" There are a number of problems. One is there aren't enough. There aren't even a hundred billion computers out there, not even ten. It's close to two billion, on all of earth. Then you MUST fit them all in a small space. The speed of light is a major limiting factor in computing. You can't go any faster than that, so if you everything miles apart, it won't work. Even more than a few cm will slow it down drastically.
By the time a computer is the size of a neuron, we'll probably understand neurons enough to start toying with them instead. Now, we are asking ourselves, what if mankind builds ourselves a super-brain, literally made of the same stuff and processes our brain is made of.
The thing that so many fail to understand, is why computers are "smart". They aren't capable of relating stuff like we can. They can't store and retrieve stuff, like we can. They can't analyze systems like we can. They are really pretty dumb. They can add, subtract, and follow really basic instructions. They can do all that much faster than we can, but you can't just add basic instructions to get complex ones.
Further, even if you could, no one wants a computer to do that. Nobody builds a computer and tells it "do whatever it feels like". Imagine some business executive trying to sell his investors on a computer program that doesn't do what you want it to. No one wants that. Sure, we may build computer programs that can analyze the stock market, and maximize investments. We can build computer programs that analyze the weather, predict strategy for warfare, etc... but they will never recommend or take action, unless the user wants them to. Even a computer that creates art still creates art as designed by the developer, for the express purpose of pleasing some targeted audience.
But because computers are evolving much faster than we do, we cant quite imagine how a future would look like if the computers have as much power as our brain has... i mean there is definately a limit to computer architecture as we know it now but its only a matter of time for the quantum computer to be a normal life thing with processing power that we cant even imagine....
and learning is literally about adding basic things to get more complex ones... when you want to learn how to play a guitar you start with the basics... and as the processing power of your brain rises to do more tasks in shorter time your skill level also improves.. and thats what the research of ai is about.. theres already a company using ai to do jobs in the middle management sector.... freelancers get tasks from this program so the company does not need the little managers to do so. meanwhile the program could also learn how to do this tasks and rate the freelancers by their work they do... the more mistakes you make the less tasks you get and thats kind of the beginning of all that you see?
+mathig nihilcek
You sir don't seem to have understandings of machine learning, neural networks, fussy logic, genetic algorithms or fitness functions.
There is no requirement for neuron-like node systems at hardware level to archieve conciousness related processes. Your tipical desktop computer is already billions of times faster for number crunching and math operations than all the humans on earth *combined*. The problem isn't the computational power so much as it is the lack of an effective all-round, general purpose, stimulus based learning algorithm.
Well concluded. He summarised what I've read from:
Tim Urban - AI blog series
Ray Kurzweil - Singularity is near
James Barrat - Our final invention
Nick Bostrom - Superintelligence
And Elon musk is executing upon it
Thing is…. 50 years is generous. We will probably see AI of this magnitude in the next 10-20 years.
Absolutely agree
It’s already here
7:31 *"This machine (A.I.) should think about a million times faster that the minds that built it, ... How can we even understand let alone constrain a mind making this sort of progress!"* - Sam Harris
Amen
Except machines can't think
@@michaels7159 our definition of what it means to 'think' is irrelevant.
@@thebloocat No...no it is not lmao.
@@michaels7159 Except the fact that its not machine
Sam Harris just has such a great way of vocalizing on this subject; great video!
H does indeed.
Sam's prediction is on the verge of coming true.
Chat gpt 4 is the beginning of the end
That last statement covered everything.
I love how the audience totally forgot of the depressing emotional response they were supposed to have by the end of the talk.
Yu
Sometimes there's things that you cannot control. If we could ally with some revolutionaries in China, Russia, the US and other countries that wanted to ethically create AI, then sure, but as it stands no one could make a change if we don't collaborate
Ben Stiller
i swear it's him
jewish descent (:
Invesigator I understand your confusion. Jewish can both mean: Following the Jewish Religion, and being of Jewish descent. I'm not too confident on the terminology but I think Jewish can both refer to race and religion.
It's why I like using the word Hebrew, like someone might call a person with the name Maclean Scottish.
Why did israel do 9/11 lol
12:14 hits a little harder after watching the Neuralink presentation a few days ago.... Sci-fi technology is already here and will continue to grow at a bewildering speed.
We are already cyborgs in some sense. We carry super computers with us at all times although the bandwidth is extremely bottlenecked because we interface with the device primarily with our hands. Once a device like Neuralink fixes the bandwidth issue, we will quite literally have the option to become a superhuman cyborg.
Progress is slowing down. Sadly
So now we should discuss this even more? With the rise of Chat GPT?
"Against the armies of Mordor, there can be no victory. We must join with him, Gandalf. We must join with Sauron."
You can't join someone you can't catch up with and especially when the gaps between the intellects gonna be who know's like million years.
@@wilhelm.reeves that's not what I was alluding to. What i meant was that the only way for AI to not destroy us is for us to become AI. We must use the technology that makes AI superior to us to replace our biochemical equivalent to it.
@@synonymous1079 would we still be people if we did that? it still sounds like ai taking over the world but in a more subtle way
@@jacobhannah2969 that's a really good question, and a really important one. It depends I suppose on how you are using the word "people". If it is referring to intelligent, sentient beings, then yes, we would be. If it is referring to the type of sentient being exhibiting distinctly human traits, then perhaps not. In the latter (and more interesting case), it then depends on what differentiates humans from other intelligent beings, and whether its preserved. It is of important note here that many evolutionarily advantageous traits are not rational. Try coming up with a rational argument for value of ones life that isn't predicated on some biological imperative. Such traits may not persist if we shake our Darwinian roots.
@@synonymous1079 AI "superior to us"? What a joke man, AI is not even real in the first place and our brains are far beyond what a computer can ever be.
4:08 she thought she was going clubbing & ended up at a TED talk about AI?
All for the love of science hehe
IKR?
I love boobs. I love science.
Were you listening to me Neo, or were you looking at the woman in the Blue dress?
( • )( • )ԅ(‾ ‾ԅ) evil robots made me do it
Honestly, the scariest thing isn't the super AI coming out as much as the current level of incompetence and corruption in our political leaders and systems which would be tasked with defining the AI safety and ethics standards.
very good ted talk . well done sir! i had never considered the idea of the dangers posed by an AI , not so much malevolent as just impartial to the lives of humans . also , killer sideboob at 4:11
was searching for this comment lol
human detected
13:15 - "I think we need something like a Manhattan Project". This is the literal phrase I have been saying about this for many months now. It's nice to hear that I'm not alone in that sentiment.
And it might be worst than the Manhattan project if a group of mad scientists would make a killing AI machines that could wipeout humanity all over the world. 😱
I think AI is a really good basketball player
"I have an dream" -Andy Burger
Almost always beat me on the Atari 2600, too.
lol
Good to see Sam implicitly referencing some excellent pioneering work done in this field - from those of us who were predicting this in the 90's and have been working on it ever since. Well-stated.
I was touched to see you bow at the end, Sam. It softens and endears my perceptions of you.
“A global pandemic” how well the speech has age
Crazy thing..it's not the last one. Can't wait for the next one!
Lol it is not that hard to predict things like that... I was aware of the issues in China's wet markets possibly leading to a pandemic, over a decade ago... when I was in high school still XD
I predict there will be a volcano eruption. Come like my comment the next time one erupts.
The engagement of Charles and Diana 👑
So.... next thing thats gunna happen is Justin Bieber presidents ??? Oh my
4:09 WTF is she wearing at a TED talk conference
She's got "mail order bride" written all over her.
ROFL, when that frame showed up I inmediatly noticed, paused the video and say to myself that there MUSt be someone down below who already noticed and made a comment, and here you are!!.
galanoth17 Russian psychologist playing a joke?
What I want to know is which of those old dudes on her side bangs her and how much it costs to maintain her a month.
I scrolled down just to see whether anyone noticed. Lol
This is a really great talk to just sum up our current time regarding the growth of technology and the seeming inevitability of artificial intelligence. At least our future will not be boring, in the next 5, 10, 20 years.. we will see the world change fundamentally, including what it means to be human.
I’ll take boring. 🙏
@@mg79277 I'll take interesting. Boring, business as usual hasn't worked out that well for us. It has been a total disaster for me. My 25 years of adult life has been a series of mental health catastrophes. AI and neuroscience could solve that. It could also solve the problem of so many people ending up in nursing homes wearing diapers and forgetting their own names. That's the status quo now, and it sucks.
Commenting here after having interacted with ChatGPT is eerie.
It's getting insanely frightening
Hello Carlosfreire8249 and Eulers_Identity,
I understand why interacting with an AI language model like myself can be a bit unsettling. However, I would like to remind you that I am just a machine learning model created by OpenAI. I do not have consciousness, emotions, or free will. I simply respond to your input based on the patterns I was trained on. While AI is becoming increasingly advanced, it is important to keep in mind the limitations and the potential dangers, as discussed in Sam Harris' lecture. But at the same time, if used responsibly, AI has the potential to improve our lives and solve some of the world's biggest problems. Thank you for your comments.
The above was manually copied from output of chatgpt. I am a real person.
@@logicturtle9838 “Trust me, I’m Bing 😊”
For those of you objecting to Sam's talk, here's a few points:
1: He had less than 15 minutes to give an overview of a vast, deep topic, and his primary goal was to raise awareness.
2: There are plenty of experts who agree with the premise of this talk. (Just as there are experts who disagree. This is how science works. That said, I have yet to see a convincing argument from the experts who disagree. Most of them are failing to see the forest for the trees, allowing personal incredulity to trump their abstract reasoning.)
2a: Just consider: Who was it that said "man would not fly for fifty years?" It was Wilbur Wright, 2 years before he and his brother invented the first airplane. There are countless examples of experts failing to grasp the implications of their own fields because they're too caught up in minutia. If it takes blood sweat and tears to get your AI to do the simplest things right now, it's difficult to see how _that_ could become super-intelligent. Experts insisted that AI could never beat the best humans at Chess, and then later at Go, and we all know how those experiments turned out.
3: Go read Nick Bostrom's book. Seriously. It will address all of your concerns and misunderstandings in excruciating detail. You have no business even forming an opinion on this subject until you've read his book. Sam is presenting the most trimmed-down, dumbed-down summary for layman audiences.
4: If you've read Nick Bostrom's book and you still think Sam is being fanciful or paranoid, read it again but pay attention this time.
Matthew Doan can I ask what those who objected were saying?
Matthew Doan does Nick Bostrom's magic book have a name? 🙄
+Cute Monster123
Yes, it's called: "Superintelligence: Paths, Dangers, Strategies".
Yes, go and read the bible, and believe. If you still don't believe then go and read it again!! Seriously, If you couldn't understand Billy Graham, go read the book again. /sarcasm off
Does this book explain the complexity of cost functions and gradient descent and their application to AI? Does it explain the difference between parallel and serial computation?
(Note: As there are some strenuous claims on his credentials on the Amazon page for the book, and seeing as he trained primarily in Psychology, not Physics/Computer Science (he did minor in these though), I am skeptical of fundamental understanding of AI rather than the philosophical remaifications of AI. In some of his published papers, he explores areas of metaphysics, not Physics itself. In reference to his "the first mathematical theory of observational selection effect", Bayes beat him by some number of years.)
I am accustomed to TED Talks providing a solution after presenting a dilemma.
He said the only thing he can do so far is sharing the message but that there should be some kind of Project Mahattan for AI, where they would discuss the matter of course.
Atomic bomb for AI. Boom.
There are world threatening problems we do not have solutions for.
The solution to everything is obviously social justice
:)
lol
Oh wow. That crowd chuckle at the start.
4:08 (lower right): when u left from home for a beach party but ended up at a Ted talk
I caught this
Sam Harris has a superpower - give him 15 minutes of your time and he'll give you a whole new way of seeing the world.
Because you will be asleep and dreaming within 10. Least charismatic guy i've ever seen.
He reeks of smug intellectualism and an inability to recognize the counterpoints to his overblown theorizing
He is not a computer scientist or a mathematician. He is a pseudo intellectual fear monger. People got pissed at him for fear mongering about Islam, so now he is fear mongering about technology. I am more qualified than he is to speak on the future of AI, and I am still in school.
People need to stop listening to media "experts", instead of listening to the people who are actually educated and/or working in those fields.
@@pythonidaepraeceptor1023 Just like the epidemiologists who were educated in the field of virology told us to stop listening to the media "experts" who were warning against labs studying gain of function?
@@pythonidaepraeceptor1023 sam harris is formally educated in neuroscience and philosophy. I think he has at least some relevant expertise as a lot of the discussion about superintelligent AI concerns philosophy and deals with comparisons to the human mind.
Try watching The A.I. Dilemma by Tristan Harris and Aza Raskin, and then wathcing this Ted Talk by Sam Harris. I bet you'll get the same eerie feeling as me as you're hearing Sam speak, that he's basically describing current events, just 6 years from that date... Certainly feels like the mothership has landed, and we are for sure not ready for it...
I did. Totally agree.
After GPT-4 and its following versions we have a solid chance to witness that "God" not in 50 years, but in less than 10 years.
The future is unknown, and the unknown is terrifying and exciting at the same time, somehow.
2:17 they laughed at the Justin Beiber joke just like the Democrats laughed at the thought of Trump becoming President. BUT you're safe Justin was born in Canada, lol
@@softan twice*
Lmao here before the angry and offended trump supporters arrive...
@@FreeSpiritPaulette I don't think trump supporters indulge in intelligence of any kind
@@beefy74 what do you mean by that?
@@DylanPiper trump supporters are dumb.
As a software developer I seriously doubt that AI can be prevented from what I call “the risk of unintended consequences” the situation is that regardless of what good intentions are out there we will indulge ourselves in AI
It's scary how true that really is. Especially when considering how many of pandora's boxes we've opened without shying away from the possible consequences. Humanity rushes on to its destruction, blind and vain. I liked your comment...
Yeah the AI control problem seems to be almost fundamentally unsolvable. Any “solution” always involves already having another solution. All we can do is try our best and hope it’s good enough
Yes then AI should be of different types for public and authorities
we need a massive spiritual psychedelic revolution to stop AI and get everyone to reset and go back to nature and value love over inteliigence but sam said it here its not all about saving humans and we are not the end of evolution. AI seems to be a force of its own and we are already transhuman with these cellphones. I wish we could all live a psychedelic burning man reality at this point as a whole but its clear thats not happening. we are not the end of all evolution and humanity is about to go through a massive ego death in order to pave way for the next idea. Its actually the ultimate spiritual awakening and yes we need to live with ourselves cause its all god from the bacteria to the AI to the stars and a fractal multiverse i suspect
human:”can we build AI without losing contr-“
AI: “nope”
Simply amazing to hear him speak
Sam really is one of the greatest thinkers of our time.
"And the people bowed and prayed to the neon god they made..."
Deep
I think they were referencing film/TV/media
Isn't that from the Sound of Silence?
They sure do.
@@cgme7076 yes
With knowledge comes responsibility. But who’s going to be held responsible when AI becomes too intelligent? People say ignorance is bliss, so why do we feel such a strong desire to know more even though it could really hurt us?
I don't know who said it (probably multiple researchers, scientists and philosphers) but I remember reading that the best way to counter the problem of creating AI more intelligent than us would be to enhance ourselves in order to keep up, i.e cybernetic implants e.t.c.
I could listen to Sam forever.