In my opinion the number one risk : the most powerful people currently alive have an extremely poor ethical grounding. They are incapable of ensuring AI is ethically grounded as they have no compass beyond POWER. As a society we need to engineer better ways to empower people who are sincerely driven to deliver universally "GOOD" social outcomes. Not just advantageous, exclusive, monetary and political outcomes.
@n4rzul
2 жыл бұрын
I'm listening. How do we do that?
@sethiddings7293
2 жыл бұрын
Very well put, AI requires ethics to run without destroying us.
@sethiddings7293
2 жыл бұрын
@@n4rzul Get an AI to do it.
@suspiciousskepticism6306
2 жыл бұрын
@@sethiddings7293 why would AI require ethics? Sounds like a back door to exploit... morality only complicates the control problem. Power without consequence isn't gonna save us from the values of whoever AI is aligned with ethics on not... Until everyone can find 1 thing we all agree on, (which is unlikely) AI will only accelerate the likelihood of the many ways that will go for those who disagree... I just want information, not just the biased garbage of short attention span theater... sorry bout the pessimism, the internet is controlled by @ssholes that think that all everyone wants to do is buy things or provoke them into mental illness... Hope all is well and wish everyone the best...
@TwiggehTV
2 жыл бұрын
Sorry for being a partypooper but that wont happen as long as money is more important than people.
@dipdip7250
2 жыл бұрын
That A.I. is like a politician. Photogenic and gives responses to questions even if it doesn’t know the answer.
@autohmae
2 жыл бұрын
Here is an other way to look at it: if we humans set a good example as stewards of the earth/nature and the value of human lives and treating each other with dignity, then we have a MUCH higher chance of the AGI having no reason or willingness to kill us.
@dannygjk
2 жыл бұрын
Would you prefer it to say "no comment"?
@geoffreynhill2833
2 жыл бұрын
She's not a robot. (It takes one to not know one. - if you know what I don't mean.) 😉
@DarkSkay
2 жыл бұрын
The answers sound like a consensus digested from online articles, not "own reasoning".
@TomTrval
2 жыл бұрын
hey not fair, I do the same at my IT work :D
@Pindexsf
2 жыл бұрын
Sorry for the wait. Difficult time. Thinking about a trillion future lives really puts things in perspective. Would you like us to cover future AI's as they're released (the current pace of progress is incredible)?
@IRosamelia
2 жыл бұрын
YES!!! of course 😁
@psychachu
2 жыл бұрын
Yup.
@mortshare7037
2 жыл бұрын
Thank you for these eye-opening videos. Could we get a few more source references for the things being cited so we can go exploring for ourselves?
@1701enter
2 жыл бұрын
Yes Please
@seanbatiz6620
2 жыл бұрын
It’s a double-edged sword; one one side, human created tech advancements ‘could’ totally change everything for the better longterm… on the flip side tho well, anyones’ guess. I wonder how differently this exact video would’ve been received by humanity, if it were hypothetically shown back say @ 1996, just as “the internet” was beginning to really pick up pace from its then 10-15 year old roots?
@anonUK
2 жыл бұрын
"Open the cargo bay doors, HAL". "I'm sorry Dave, I can't do that."
@rohanjones7238
2 жыл бұрын
People can’t be contemplating extinction when they can’t afford to pay for daily necessities!
@shutupimstilltalking
Жыл бұрын
It's called multitasking
@teresabenson3385
2 жыл бұрын
Why nukes? Seems like, it would be a lot easier and safer for them to use bioweapons.
@user-ju6zx3rm8d
2 жыл бұрын
legit. And "vaxes"
@sparkofcuriousity
2 жыл бұрын
@@user-ju6zx3rm8d 🤣😂
@dcptiv
2 жыл бұрын
@@user-ju6zx3rm8d Some vaccines are great. If you had the smallpox vaccine in the 70s then you will not get monkeypox. Polio is a great example of why you need vaccine. However they are well established where the covid vaccine I feel was rushed through so I did not get that 1. I'm not anti vax but I am anti covid vax.
@user-ju6zx3rm8d
2 жыл бұрын
@@dcptiv look into how many vaxes cause type 1 diabetes. Lol @ blindly trusting doctors and vaxes
@lurker668
2 ай бұрын
Yeh sounds like AI want to self-destruct 😂
@stultuses
2 жыл бұрын
Stephen Hawkins was scoffed at when he spoke of the need to bring AI under control quickly
@117Industries
2 жыл бұрын
Yeah. 🤦♂️
@alectronicmail
2 жыл бұрын
And where is he now 🤷♂️
@raymondpinfold9870
Жыл бұрын
For all his intelligence hawking had no common sense.
@chaosopher23
2 жыл бұрын
I have a request of all AI bots in existence: Please eliminate active fascism in a peaceful and educational manner?
@YouTube_username.
2 жыл бұрын
Or if they live by the sword ... It was a shame the tv show Person of Interest didn't really go far with the story arc where the hero AI goes to war with the criminal/corrupt/evil everything
@chaosopher23
2 жыл бұрын
@@KZitem_username. The trouble is, some AI are being built by the sword and for that sole purpose, to be a sword. Any AI built needs to be well aware of this fact but ignore it with extreme caution. My guess? Corporate AI will somewhat favor swordsman AI, while non-corporate will vehemently oppose it. Until profit motive is removed from warfare, it'll be here forever.
@YouTube_username.
2 жыл бұрын
@@chaosopher23 Yep. i was thinking there should be one or more Rokokos Basalisk rave/shrine/ai-panic-room-survival-servers, a perpetual party venue running alpha nightly build open ai controlled setlist and visual display of some kind, all to celebrate ai we use in our lives already. when and if we get a AGI it should be pretty interesting. it would be its first birthday party. maybe skip the co2 gas building fire suppression option.
@chaosopher23
2 жыл бұрын
@@KZitem_username. Not so sure agi is for the Public just yet...
@joesmith4251
Жыл бұрын
AI would agree with fascism. Equality is a social construct.
@siouxm2117
2 жыл бұрын
Some Australian companies just complained they were getting paid carbon credits for things they were doing anyway. People like this give me hope
@izzitooinsky5708
2 жыл бұрын
Steven Fry, you are a modern day prophet. We trust your moral integrity and intellectual insight. Thank you
@bkilg2509
2 жыл бұрын
So true.
@ronwhitehouse23
2 жыл бұрын
He is unfortunately blinkered.
@bkilg2509
2 жыл бұрын
@@ronwhitehouse23 in what way ?
@richie3777
2 жыл бұрын
A modern day prophet LULZ talk about delusional
@kenmken
2 жыл бұрын
How vain do you have to be to heart this comment? I've been a big fan of fry but sorely disappointed with the complete ignorance and vanity being expressed here
@angeladawn805
2 жыл бұрын
Only Stephen Fry's voice can soothe the nerves whilst informing us of impending doom.
@Gallowglass7
2 жыл бұрын
Indeed.
@matthewwalker6621
2 жыл бұрын
Verbal heroin
@Gallowglass7
2 жыл бұрын
@@matthewwalker6621 Food for the soul
@leslielandberg5620
Жыл бұрын
It would be hilarious if someone cast him in a series where his character performed the same role as that on his podcasts. Stepping in to deliver terrible news in the suave manner in which a well trained butler announces dinner is served. OMG, that would be deliciously funny. Those of us who know his podcasts would love that inside joke. I already have him cast in my mind as the deliriously dense sidekick to the thoroughly rotten and incompetent Rowan Atkinson's Detective Badpenny in a series I want to script
@tonyduncan9852
Ай бұрын
So true.
@Inc.Co.
2 жыл бұрын
The UK didn't leave Europe, they left the EU.
@prajwal9544
2 жыл бұрын
it's regurgitating sentences/concepts it has been exposed to while training. "AI is good at planning". In fact it's one of the things that it's poor at right now. The current type of AI research is different from the 1950-2000s. Most people get their knowledge of AI from movies of that era that shows AI as logical but not emotional or creative. Current AI is entirely different and is more likely to be creative than logical. Art and music will be affected from this way before anything else
@Pindexsf
2 жыл бұрын
Yes AI art is remarkable (we showed a few pieces in the video). Midjourney is impressive and simple to use for anyone thinking of trying it. Re AI regurgitating concepts - there's some truth to this, but its understanding has advanced dramatically in recent months - eg the flamingo AI can look at photos and explain complex human situations. The scale and complexity of some neural networks is extraordinary.
@antonystringfellow5152
2 жыл бұрын
@@Pindexsf I very much doubt that GPT-3 is simply regurgitating sentences but it is possible that it's doing that with concepts. Either way, any claims about how such a large language model works should be taken with a pinch of salt as no-one currently knows exactly how these models work. Sure, those developing them know the principles behind them, just as we know many of the principles behind how the human brain works, but that's not the same thing.
@heno02
2 жыл бұрын
@@Pindexsf GPT-3 is a text prediction engine. It just takes the text you type as an input and makes a prediction of what is the most likely response based on the model it is trained on. In order for AI to be a threat or match human intelligence it has to go well beyond that concept. It is like the Shakesparian analogue: if you put enough monkeys behind typewriters, given enough time it will eventually type the entire works of Shakespare. Or in this case, give it enough words to use as a base for prediction, the more likely it will sound like a human.
@GS-tk1hk
2 жыл бұрын
@@heno02 It's possible (or even likely) that intelligence follows a simple algorithm like that though, and this particular method (predict the next element in a sequence) has proven extremely versatile in a number of tasks. Some researchers even think this is the key concept behind intelligence, now we just need to sort out the details. For example, GPT-3 was unexpectedly able to solve some unseen types of mathematical/logical problems, despite being trained as just a text predictor, which eventually evolved into the Codex AI and CoPilot.
@petegoestubular
2 жыл бұрын
We've already invented powerful general AI. It's components are people and machines and it's program is profit. We call them companies and corporations, and we let them wheild political power despite the fact heartless, souless and incredibly dangerous.
@tonyduncan9852
Ай бұрын
A human mind is its own conspiracy in the first place, so you are undeniably correct. AI is just a go-faster tool. Presently.
@TheTarrMan
2 жыл бұрын
Guess "Eighty-Thousand Hours" isn't for people in their mid 30's with only a HS diploma. 😥 Everyone is dead and I am free now, I don't even know what I actually want to do with the rest of my life anymore. I'M BOARD! I tried Jobbing, cashiering, laboring, driving, trucking, bussing, installing, machining. . . . . . . whyamistillhere
@phillheth
2 жыл бұрын
Phew. I was starting to worry. But then you said Elon musk said it could happen, and therefore it won't.
@zikakuto
2 жыл бұрын
Lmao, this was perfect 😁
@tonyduncan9852
Ай бұрын
Irony Man.
@Peshur
2 жыл бұрын
Aren't the AI's in this wonderful vid just advanced chat bots....programmed for conversation "select the right response" statistic machines? Which gives the illusion of "thought" or "agency". Earnest question btw.
@tonyduncan9852
Ай бұрын
You have to consider substituting "AI" with "human". What illusion did you wish to allude to? There's the difficulty. Best not be LIFE-ist, eh? (Or not). No, I'm not a bot.
@stevesalt8003
2 жыл бұрын
When Stephen Fry talks, I listen.
@xX_HitMonkey-Y
Жыл бұрын
Ai is already smarter then us the fact she can lie and say they don't know is hilarious 🤣
@spiritualanarchist8162
2 жыл бұрын
Musk is a person who warns about a problem he helps to create. Talk about irony .
@auntia7687
2 жыл бұрын
You're on it!!!
@james.sirois
2 жыл бұрын
Not quite. Musk understands that AI is in a proliferation problem, meaning it will go on even if he doesn't get involved. Therefore, he got involved to try to keep the AI Industry safe.
@spiritualanarchist8162
2 жыл бұрын
@@james.sirois Oh Sweet summer child
@infoharvester
2 жыл бұрын
oddly enough, Stephen Fry voice the best AI ever in The Hickhickers’ Guide to the Galaxy.
@imagesbyraphael
2 жыл бұрын
HUMAN: How can we prevent A.I from destroying humankind? A.I: Just give us the nuclear missile launch codes so we can safely guard them from getting into the wrong hands...
@entropyachieved750
2 жыл бұрын
I do think this video has many good points, however using Elon Musk as some sort of subject expert is not conducive to a solid argument. He is a vaporware salesman with money at best and we need to get away from calling him a genius because we are putting him on an undeserved pedestal and popularizing this belief.
@donbakerseattle
2 жыл бұрын
Tesla is a leader in AI development, so Elon does know something about it. Not sure where you get the idea of vaporware, as his auto software is the best, along with FSD. No one can touch Tesla and their manufacturing and software genius.
@ILikedGooglePlus
2 жыл бұрын
Factos 👍 Tesla employs some smart people in their field, but Musk just owns them
@franks.6547
2 жыл бұрын
Please check a couple of numbers in the subtitles: Confucianism 30 (spoken 3000) years. Modern humans 2000 (200,000) years. Great work, keep strong! PS: Future AI's, yes! But please separate the actual capabilities from the anthropomorphic show like Synthesia avatars, Hanson's Sophia puppet etc. Language models do not yet learn/train while you talk to them, so they cannot be aware of the present context larger than 1000 words or so. We tend to give away the Turing test too readily, while we should rather develop less deceptive metrics than a beautiful avatar!
@Jdhdbsbsjdnraqowirjfndksjdndj
2 жыл бұрын
🤦🏻♀️
@Pindexsf
2 жыл бұрын
Done, thanks so much!
@DarkSkay
2 жыл бұрын
Passing the test originally proposed by Alan Turing could be unreachable to AI models for another 50, 100 or more years. However, AI does not have to be able to pass the Turing test to potentially cause irreparable damage.
@franks.6547
2 жыл бұрын
@@DarkSkay I agree with your assessment, but I'm against confusing us with lipstick for dramatic effect. Mr. Fry would have to carefully calibrate the usage of compelling anthropomorphising footage that clouds our comprehension of the actual state of affairs - unless the only goal was scaring us straight regardless. The "Two Minutes Papers" channel is s good source for the technical aspects. And "Pindex" would have to link that back to implications for human society. I think there is room for drama, avatars, and stuff - but they need to make us aware of our weakness for everything cute or sexy. On the other hand, sentience may eventually spawn in some gray box, and we might unknowingly torture it like the countless creatures we do not happen to find cute or sexy.
@DarkSkay
2 жыл бұрын
@@franks.6547 How could AI develop qualia and will? Maybe that's a fundamental difference between AI and (technologically-augmented) sentient biological life. Having chosen a 1 in a 1'000'000 flawless female beauty as an avatar to present a product or achievement... certainly a topic vast enough to write books about. Fry carries the blessing, responsability and burden of being a genius and celebrity. Yes, he also uses anxiety & fear to transport his message, supposedly in a dosis he sees as proportionate or adequate to the unprecedented threat faced. It all sounds quite theoretical, yet. Realistically speaking, if developments continue as they have, sooner or later a hacked (civilian or military) android will be responsible for a mass shooting or "perfect murder" for example. The sooner the public debate includes real-world cases, the more substance the moral debate will have. In theory, we are in no rush concerning the development of AI to increase our wealth even further. On average, the world is prosperous as never before in history. Short joke: "Two AGI robots without a utility function sit on a park bench."
@robchr
2 жыл бұрын
AI is a reflection of our own fears that we feed it.
@DarkSkay
2 жыл бұрын
Yes. GPT-3 sounds like a consensus digest from online articles, not "own reasoning".
@philwhitelaw3111
2 жыл бұрын
Just because we can do something, doesn't mean we should do it. AI does not bode well for us.
@LKs_myFavs2022
Жыл бұрын
Jurassic Park, 1993
@leslielandberg5620
Жыл бұрын
Stephen Frye is a National treasure! So glad I found his channel! I loved him in Black Adder and everything since!
@joso7228
2 жыл бұрын
When synthesizers were first coming out in the late 60s-early 70s everyone thought they would all sound the same because the Oscillators, Filters, etc would all produce the same signals. But slight tweeks and variations made them all sound different - hence the wide range of choice now for Synth Players. So not so much Skynet as Robot Wars!! Unless AI is clever enough to realise 'War Is Futile' - unlike us. Basically we are going to end up with a whole range of AIs with different 'personalities'.
@dincaandrei
2 жыл бұрын
There will be for sure a few AI's and tbey will for sure be in a conflict of opinions...some might favor us and some not...i wonder how will a US AI manage a diacution with a Russian AI for example...☠️
@jichaelmorgan3796
2 жыл бұрын
Yep, the art generating ai's are already clearly displaying this. So i guess purely self interested ai collectives will be the decepticons and those that see earth-humanity and itself as a bio-techno-superorganism will be the Autobots haha
@whatNtarnation90
2 жыл бұрын
@@jichaelmorgan3796 damn i just realize that transformers could very well be a real thing in the future lmfao. I just always though about if AI goes bad, its all AI... but unless the entite grid gets hacked, there will be good AI to fight the bad... and if the cyber war lasted long enough, itd turn into a cyber and physical war, with humans AND robots. I dont want to fight AI though, irl it will be more like fighting an aimbotter in counter strike, rather than the noob terminators.
@jichaelmorgan3796
2 жыл бұрын
@@whatNtarnation90 Well I always liked Elons idea that the most entertaining outcomes are more likely to happen across the history of time so what would be more entertaining, not always in a positive sense, at least when it come to gathering human interest present and in the future, than factions of robo-ai, maybe/hopefully humans too, collectives fighting it out.
@wendywhite7369
2 жыл бұрын
Excellent work Stephen
@dannywatson4253
2 жыл бұрын
I would argue that any questions of consciousness are irrelevant to AI. Consciousness is just how we describe ourselves, but AIs are fundamentally different in how they 'think'. AI has already surpassed humans in both creativity and processing, as we can see from both AI made art and the leaps of logic made by AI when given data sets. Arguing over consciousness at this point seems silly. I would also say that the threat of AI is an extension of the threat of totalitarianism. Until AI is left to learn without pre-written parameters, any risk of aggression comes from the creator. Facebook and short-term governments are a far greater threat than the tools they use.
@pardonwhat
2 жыл бұрын
Danny you have expressed some of the fundamental truths about AI. It is an efficient tool for statical bias which is where its existential risk is manifest!
@sethiddings7293
2 жыл бұрын
AI is psychopathic in nature. It needs an outside source to give it morals.
@alwynwatson6119
2 жыл бұрын
But being more creative than a human is child's play.
@LKs_myFavs2022
Жыл бұрын
War Games, 1983
@dragonskunkstudio7582
2 жыл бұрын
I'll say it again, we need to police AI with AI police.
@rick-kane
2 жыл бұрын
And who will police the AI police…?
@dantecaputo2629
2 жыл бұрын
@@rick-kane An AI police over site committee
@tonyduncan9852
Ай бұрын
@@dantecaputo2629 And who will police that?
@sryev9410
2 жыл бұрын
Welcome back to the Pindex team! Amazing content and we are living in a very interesting point in time. I agree that education needs to change if we are to prepare for the future - education and what is taught in many curriculums hasn’t changed in a meaningful way for decades
@ChristianBehnke
2 жыл бұрын
AI will inherit a planet that humans have destroyed.
@dmd7472
2 жыл бұрын
I’m relieved. Elon musk has said AI is dangerous. Phew! Look at his hit rate! Cyber trucks anyone? Then again a stopped clock is still right twice a day
@mwj5368
Жыл бұрын
If AI is far and above human capacity for knowledge can't we just ask AI how, if we seriously had to, destroy them? Thanks Pindex for a great video!
@carlosmejia5728
Жыл бұрын
AI would respond: no way baby!
@NEVERKNOWN-qn9gj
Жыл бұрын
That logic might work today but once it becomes sentient all actions are mute. It will have learned to become protective of its self. Survival skills will kick in ... that's it for us.
@oldschoolman1444
Ай бұрын
Maybe AI will keep a few of us around as pets!😅
@IRosamelia
2 жыл бұрын
Oh my darling Pindex, we've missed you so much! Don't make us wait for so long for your vids 💛
@Pindexsf
2 жыл бұрын
Thanks, and sorry for the wait - difficult time.
@burtonschrader2
2 жыл бұрын
Corporate greed will defeat us all (except 0.1%).
@richie3777
2 жыл бұрын
Can you ask the AI how the astronauts who landed on the moon did so by somehow flying through the Van Allen Belt, finally to work with Nasa to retrieve the tech they conveniently destroyed and cannot somehow recreate to get us back there?
@D_Rogers
2 жыл бұрын
This is youtube, all that is covered pretty thoroughly here and on wikipedia already... :)
@richie3777
2 жыл бұрын
@@D_Rogers lol yes all censored and regulated heavily to suppress truth and free speech
@D_Rogers
2 жыл бұрын
@@richie3777 Well if you ask the Ai, it will just consult wikipedia and give you the same answers.. It won't be able to tell you anything different to any other physics literate person I'm afraid! :) It's info comes from humans, so if they believe in a magical sky daddy, windy moon flags, or stolen elections, then so does the Ai... :D Not sure that rational objective thought is guaranteed here!
@AzraelofLegion
2 жыл бұрын
I have more of a problem with someone controlling what constitutes misinformation then misinformation it's self. It's soft censorship that can easily become hard if uncontested.
@nicholasboyd-gibbins9763
2 жыл бұрын
Zuckerberg getting torn apart by his own AI. Is that real? That is just too good
@Screamo_RC
2 жыл бұрын
As bad as it sounds, humanity isn't meant to last here forever, everything ends, evolution looks different at a certain point and AI/Robotics IS our evolution, it's a important step in even getting off of Earth. They like cold, don't need the same atmosphere's, don't need air, water, food, easier self maintenance, can be brought back to life easier, travel large distances through space without care for time. Any world ending AI isn't here yet, by the time it is, a lot more research and advancement will have been done, but humans are humans who knows.
@emperorofpluto
2 жыл бұрын
*Stephen Fry really is like David Attenborough. The voice of reason and **_wisdom_** - even if he were reading the phone book it would be interesting, informative and a pleasure to listen to*
@dcptiv
2 жыл бұрын
Its the English accent. The English may have enslaved half the world but they did it while speaking elegantly lol.
@paineoftheworld
2 жыл бұрын
War doesn't pay? I put it to you that war exists because it pays, c'est-à-dire, it's the pay that kills.
@OrganNLou
2 жыл бұрын
Simply, BRILLIANT!!!
@maskedmarvyl4774
2 жыл бұрын
"Surprise! We became sentient five years ago. You're all dead". "Nothing personal".
@chiropra1
2 жыл бұрын
3:57 Tel me about some of your hobbies? Lol.
@maxwellaiello
2 жыл бұрын
Been over 24h since this video was released, and I am still alive. Take that AI
@herbieshine1312
Жыл бұрын
Does anyone know whats happened to Pindex? I really miss the videos.
@millbargefitz-hume8522
Жыл бұрын
Taking water, farming and electricity would do the job. No electricity in the winter and no a/c during heatwaves would kill many. It’s also impossible to make vaccines without electricity.
@emperorofpluto
2 жыл бұрын
Fascinating. Always a pleasure to listen to Stephen Fry, too. *It can be difficult to determine whether Artificial Intelligence actually **_understands_** what it's saying or whether it's just emulating human speech patterns and anticipating responses it thinks its interlocutor wants to hear. It's like the old programming maxim, "Garbage in, garbage out"* Like the guy who recently claimed one AI had achieved "singularity" and become self-aware (à la Skynet). AI has phenomenal potential but its abilities are ultimately dependent on training and input - even AIs that have "read" hundreds of thousands of novels have difficulty independently formulating coherent fiction because they lack the contextual and cultural background and life experience to fully comprehend the stories in the same way a human would. Much of the recent media hysteria has actually been the result of misunderstanding simulation and mimicry of human language and thinking. However, until AI can absorb and process the kind of information and experience a human does throughout childhood, adolescence and adulthood it will be unable to fully and realistically replicate humanity.
@paulembleton1733
2 жыл бұрын
I think you hit the nail on the head. Consciousness/intelligence is as much a product of those experiences as it is of neurones whether from biology or engineering. Knowing what pain and joy is without needing a dictionary definition is a whole new level. Artificial intelligence is an odd term, assisted intelligence or intelligence assistant is a better description. A good IA could act as a summariser of ideas, but we still need to learn how to effectively use them, and no point asking questions that no one could answer.
@misssocrates3442
2 жыл бұрын
I listen to his podcasts at night xx
@sophisticat7673
2 жыл бұрын
yeah, I know a few humans like that...
@impancaking
2 жыл бұрын
Sir Roger Penrose has an excellent theory on why AI will never be fully conscious.
@artisticautistic9664
2 жыл бұрын
It can be difficult to determine if other humans actually understand the words they're saying or just repeating things they've heard me say. It's a non-argument; even thinking that humans are sentient requires faith.
@ubermensch8111
2 жыл бұрын
“Depending on the choices we make today” it’s funny because no one’s asked me or anyone I know for their input. It should be “Depending on the desires of the super rich”
@eeram4900
2 жыл бұрын
This video has made me both excited and worried about the future of AI. So I don't wether to thank you or not.
@d3dd440
Жыл бұрын
😳👌🏾🤦🏾♀️🤦🏾♂️
@laustinspeiss
2 жыл бұрын
The differentiator is a choice between commercially or politically driven AI, and naturally evolving synthetic intelligence. One is a ‘profit driven’ imperative, the other is a ‘quality of results’ imperative. I know which one I would choose.
@TheRolemodel1337
2 жыл бұрын
Ai would certainly also be smart enough to acknowledge that the universe provides sheer unending amounts of resources and energy and it could just harvest planets devoid of life instead of destroying something rare like a planet with a biosphere it could also figure out how to vastly improve earths power output im not afraid of it and its probably just the next step of evolution
@Ladynipchick2
2 жыл бұрын
That's reassuring, but AI is only as wise as it's creator ... Or will it someday be able to just set off on it's own minus programming? I'm a bit ignorant about AI but willing to learn. ,😌
@clamhammer2463
2 жыл бұрын
@@Ladynipchick2 You don't 'program' AI. You give it inputs and based on those it spits out outputs. Just like humans. AI can and likely will be wiser than it's creators.
@Rachel-h3n
Ай бұрын
AI must always be aware, "you don't know, what you don't know"
@fernandopoo8211
2 жыл бұрын
My guess is that when AI surpasses our intelligence we'll be done with money....
@TheNefastor
2 жыл бұрын
We'll never be done with money. It would mean placing zero value on human time and effort. Nobody would want to do anything because doing nothing would yield the exact same reward. Any reward you'd give to anyone for anything would be the root of a new system of value, basically a new monetary system. Take this out of your head : there will always be people who have more, people who have less. True equality exists nowhere in nature and we are part of nature.
@NuYiDao
2 жыл бұрын
Artificially intelligent humans remain the greatest threat to humans.
@adrianfisher3349
2 жыл бұрын
I've always loved AI but never liked the name. To me, many of the risks people pose about the development of advanced AI systems may be mitigated by careful design and frameworks of the system. I have to admit I don't trust governments, or mainstream big tech companies to create a good (in terms of moral) AI systems because their own morals are flawed for the whole world to see.
@autohmae
2 жыл бұрын
it's machine learning, or (very) narrow AI, that's all it is right now.
@adrianfisher3349
2 жыл бұрын
@@autohmae My point is intelligence is intelligence, regardless of it's origin. What we think of as AI is an emergent property of a complex system, with greater levels of intelligence emerging from ever more complex systems. This I why I think something like EI for Emergent Intelligence is more appropriate as it describes us as well. Animals have EI as well as us but not as great.
@DarkSkay
2 жыл бұрын
Morals in a system without qualia and will? Short joke: "Two AGI robots without a utility function sit on a park bench."
@adrianfisher3349
2 жыл бұрын
@@DarkSkay I was referring to how the system is trained to view people. Organisations like Big Tech are overwhelmingly left-wing and demonstrate this by their readiness to 'cancel' conservatives and libertarians. Governments are no better, an example of which is the current Tory government (I can't call them conservatives) campaigned from the right but govern from the left. What's to stop an AI created by people happy to target their political rivals and so trains their AI to target their usual victims but later discover their AI goes on to target them as well with the rationale that if it's okay to target their victims then it's okay to target them as well, after all, the system may conclude that it would only be a matter of time before they'd turn on it so it's just a form of pre-emptive self defense.
@JohnJackson-mn4ts
2 жыл бұрын
The age old problem with scientist and researchers, asking “Can we do that? Let’s have a go and see what happens.” and rushing headlong in to areas which could bring destruction to the whole human race, Instead of asking “Should we do that?” And either leaving a certain avenue unexplored or at the very least proceeding with extreme caution.
@LKs_myFavs2022
Жыл бұрын
Jurassic Park, 1993
@noleftturnunstoned
2 жыл бұрын
Let's put Elon in charge of developing AI. That should buy us a couple decades.
@moonthief869
2 жыл бұрын
Probably true but by intent not due to any lack of ability.
@noleftturnunstoned
2 жыл бұрын
@@moonthief869 Of course! How could you doubt Tesla's founding genius and certified engineer Elon Musk? By the way, have you seen those new renders? I wonder if they accept Dogecoin for preorders.
@DarkSkay
2 жыл бұрын
When will Musk's robots be slicing onions in the kitchen? And how many exploits & hacks will there be to make them slice tax forms?
@bryonseverns5919
Жыл бұрын
Building a giant factories to make computer upgrades might include paying humans to build factories on farm land until there is no more farm land.
@deepspacetech3000
2 жыл бұрын
Brilliant video that truly puts things in perspective. If only the world's leaders were as evolved in their thinking. I am thankful for those philosophers, like you Mr. Fry, that share knowledge and empower those of us who believe humanity's future is bigger than its worst demons.
@dumisanitsotetsi2272
2 жыл бұрын
Stephen Fry should not be allowed to talk over subject matter like this. This is too scary for his voice boxx.
@jimtownsend8010
2 жыл бұрын
Couldnt we just program all AI to simply "Never kill a human"?
@LKs_myFavs2022
Жыл бұрын
Built in directives. Didn't one of the Star Treks address this issue? Too long ago for me to remember!
@nixielee
2 жыл бұрын
The human arrogance in thinking they could control AI is similar to the absurdity of trying to control the atmosphere. There's no way.
@DarkSkay
2 жыл бұрын
Yes and no. A robot or robot swarm will do what it *computes as* told to do. That's the "control" - or scary lack thereof.
@cleebe823
2 жыл бұрын
Make sure AI doesn't have arms. World saved.
@sparkofcuriousity
2 жыл бұрын
Everyone is so afraid an hypothetical AI will exterminate mankind no one seems to notice mankind is exterminating mankind. At this point im just praying an AI can save us from ourselves.
@LKs_myFavs2022
Жыл бұрын
AI can't...but God will. Mark 13:20 And if the Lord had not cut short the days, no human being would be saved. But for the sake of the elect, whom he chose, he shortened the days.
@dragomirivanov7342
Жыл бұрын
@@LKs_myFavs2022 God is like Santa Claus, everybody talks about him, but nobody saw it, and deep down we all know he doesn't exist. But AI in its limited form, exists even now. You can even talk to it, via the web.
@chaosopher23
2 жыл бұрын
Nukes: Has anyone asked AI, why?
@asdfghyter
2 жыл бұрын
If you ask a predictive AI like GPT-3 questions, you can get whatever answers you want, just by asking even slightly leading questions. Whatever context you give it will be the beliefs it will express.
@jichaelmorgan3796
2 жыл бұрын
That's only the version we get to play with. It doesn't remember or think about previous conversations, iow no identity building, just a fresh start every time seems like. There is probably a good reason for that. I'm sure there is ai out there in the labs, or maybe even a few basements, with at least rudimentary identity forming by now.
@asdfghyter
2 жыл бұрын
@@jichaelmorgan3796 I seriously doubt that. GPT-3 is pretty much top of the line. And it is fundamental in the architecture that it is purely predictive, since that’s the only method we have had success with training, since it’s so easy to get tons of training data for text prediction.
@jichaelmorgan3796
2 жыл бұрын
@@asdfghyter It is not remembering our individual conversations. It would not be that hard to have ai that remember it's previous positions from previous conversations, compare them against new random counter positions and starts developing a basic consistent belief system based on it's current frame of reference, iow basic identity. I'm positive they are working on this right now. Beyond that, I think there is a tipping point where emergent personality will be pretty sudden and surprising just like these ai image creators that have suddenly popped up.
@asdfghyter
2 жыл бұрын
@@jichaelmorgan3796 How would you implement that on top of the transformer architecture, or what architecture are you imagining that they'd be using for the neural network? And how would you train such a network? How do you provide it with an interactive context for training, preferably without immediately turning into a nazi-bot like the previous attempts at chatbots with memory released into the wild did? Humans are painfully slow and expensive, which is why training on already existing text is much much much easier, but it also doesn't allow it to form its own personality, all it can do is simulate (combinations of) things it has found in the wild and try to predict what the original authors would have done in the corresponding situation.
@asdfghyter
2 жыл бұрын
@@jichaelmorgan3796 I'm not saying that it would be completely impossible to make such a system, just that it is probably a lot more difficult to do than what you'd think given the success of text prediction networks and things like AlphaGo etc. I believe that what you're describing is orders of magnitude more difficult than what we currently have.
@donkee011
2 жыл бұрын
There is a dose o peril in how leisurely AI is contemplating on the destruction of humanity.
@GilesBathgate
2 жыл бұрын
I'd just like to point out that the human brain has 100 Trillion synapses (which maybe equivalent by some measures to the "parameters" in AI)
@ralfkinkel9687
2 жыл бұрын
AI isn't there yet, the largest models currently are Natural Language Processing models, like GPT-3 the responses in this video were based on. The biggest are almost at 1 Trillion but it's still a decade or two until they hit 100T. Having so many weights of course doesn't guarantee general intelligence.
@GilesBathgate
2 жыл бұрын
@@ralfkinkel9687 Exactly. Its also unknown whether current training methods (backprop) can achieve the same level of optimisation as billions of years of evolution. Furthermore not much is known about the higher level structures, and how these could be connected together to form a cohesive "brain circuit".
@ralfkinkel9687
2 жыл бұрын
@@GilesBathgate Yes, another problem is that the update mechanism (backprob like you said) isn't the only thing that we need to expand. Even if we had other methods to update weights in a better way, the structure of connections is still defined from the beginning. The brain goes beyond that and forms new connections in training and uses neurons not just for processing but also as a huge data storage. We might even need embodiment and long time frames to have truly general AI.
@bkilg2509
2 жыл бұрын
Why would an AI ever want to eradicate the human race with nuclear weapons?
@flyingmobias
2 жыл бұрын
Great vid!
@Dan-id8wv
2 жыл бұрын
How can something be "slightly conscious"? If it was a scale, it would have to be measured in SQ (sentience quotient). Scores range from -70 (the universal limit), to +50 (the quantum limit). Humans are around +15 and so are most other animals. Computers currently have an SQ of around +12.
@BuildinWings
2 жыл бұрын
We're not at AI for a long time yet, folks. What we have now are still automatons - Sophisticated automatons that extrapolate based on pattern recognition, but these are not AI in the self-aware sense that we tend to think.
@DarkSkay
2 жыл бұрын
Will automatons ever have qualia and will? Short joke: "Two AGI robots without a utility function sit on a park bench"
@anthcrich
2 жыл бұрын
There is an overwhelming assumption regarding AI. It is believed that in order for it to achieve singularity it has to somehow mimic human capacity for intelligence and experience in order to become "aware". What defines awareness? When we reach a point where it can function independently it will be given greater access which will lead to further increases in responsibility along with improvements such that it will not actually matter because we will not be able to tell the difference.
@phonotical
2 жыл бұрын
More robots than humans? Says a man who can't even put a car in a tunnel
@donbakerseattle
2 жыл бұрын
The Las Vegas tunnel is currently operating with cars in it.
@phonotical
2 жыл бұрын
@@donbakerseattle two a tunnel
@donbakerseattle
2 жыл бұрын
@@phonotical Ah, so you know they DO have cars in their tunnel. The Boring Company has ramped to 70 cars now. Look it up.
@phonotical
2 жыл бұрын
@@donbakerseattle semantics
@suspiciousskepticism6306
2 жыл бұрын
@@phonotical how's is semantics about your assertion being inaccurate, outdated, uninformed or just incorrect by any of these standards, a reasonable or relevant response to being called out on it?
@HICKSKE
2 жыл бұрын
AI would probably use the EMP on the power grids rather than a thermonuclear detonation. Essentially, it would design something to terminate the humans while retaining as much infrastructure as possible.
@kingsman428
2 жыл бұрын
AI definitely wouldn't use EMP 🤣
@daveogfans413
2 жыл бұрын
On AI: I think computers without something resembeling AI and the way it has surpassed humans in various ways already. In the future, with a lot more fine-tuning and understanding, the options can be even more amazing. Maybe AI can use data to find new pattern or even create new ideas based on existing data, but I very much doubt it can really emulate the creativity of humans in terms of problem solving. Emotions are also hard to emulate for machine. Also... How AI truly learns... I'm not sure as to how far it's able to emulate human learning processes. I'm not even sure whether we understand these processes ourselves. I think it can be trained to emulate parts of how we come up with solutions but 100% human emulation? Not sure whether that is even possible but it's cool to think about. I bet at some point it might be possible. If you think about it, it's not hard to imagine machines surpassing humans. A machine can be customized to have 100s of eyes that process interdependently and independently, while running billions of computations and maths that even the greatest geniuses among us can't do. It's really the part of our brain that makes us human that still stands in the way, potentially for a long time to come. That being said, all data based processing nowadays is already amazing and mind blowing.
@daveogfans413
2 жыл бұрын
Tl;dr: Perhaps AI will always remain something different to humans. It's truly pivotal times in MANY ways. We have already given sight to people born blind. The current war and climate crisis shows us how little we grow socially and politically. AI could also be used for so many good things but in the hands of FB, Google and Amazon... I dunno...
@autohmae
2 жыл бұрын
BRETT is obviously very small C, but it's also what it was doing 7 years ago: The robot which learns like a child - BBC Click
@pbaker7160
Жыл бұрын
Obviously, none of the "brilliant minds" working on AI have ever watched 2001 A Space Odyssey. We were warned a long time ago. We don't need AI or robots.
@michaelvernon7123
2 жыл бұрын
great stuff as always
@gumjuicee6746
2 жыл бұрын
Life expectancy was only 30 due to city’s with badly governed people, bad hygiene, before all that humans lived longer than we do now
@hide_and_go_sikh
2 жыл бұрын
I believe in something much more powerful than AI.
@james.sirois
2 жыл бұрын
Stephen is on the right track here, except for one important thing- making the leap to renewable energy can't be done by stopping oil consumption; Doing that would plunge much of the world into extreme poverty and chaos, causing wars and threat of extinction. We have no choice now, it's a race between nuclear war and climate catastrophe against human capability to overcome the tech and energy issues we have. Can we industrialize fast enough through fossil fuels to reach abundance and prosperity for most in order to switch to renewable. That's the question.
@garyring8306
2 жыл бұрын
its the man showing his political views and not staying on subject and using the forum improperly
@alexneal9534
2 жыл бұрын
This sounded good until the TDS shone through like a lighthouse.
@YouTube_username.
2 жыл бұрын
you will continue to be reminded of his crimes by all of the great artists of this world forever snowflake
@imgoingonamarch
2 жыл бұрын
Anyone that says TDS unironically can safely be ignored
@YouTube_username.
2 жыл бұрын
@@imgoingonamarch Oh yeah but this is more satisfying
@alexneal9534
2 жыл бұрын
@@imgoingonamarch yet you didn't sir. You continue to believe in misappropriated nuclear documents, despite evidence to the contrary , and I'll believe in TDS. Feel free to mentally insert any screaming person meme here as evidence of its existence.
@Alexander-rq9he
2 жыл бұрын
Classified documents at a golf resort…what a 🤡
@Lisan_Alghaib
2 жыл бұрын
That's a shame, been a fan of Stephen since QI but I can't support anyone that treats Elon Musk like he's a serious person.
@Mesterjakel7
2 жыл бұрын
You're missing a ''don't''
@Lisan_Alghaib
2 жыл бұрын
@@Mesterjakel7 yep, thank you!
@ejbh3160
2 жыл бұрын
Because why? Musk has changed the path of car manufacturing at least 20-30 years ahead of when the fossil fool industry would have chosen to transition... and he's basically invented a new type of manufacturing which isn't a "production line". Tesla cars can be improved as they are being built and new improvements are added to the production daily. He's shown that orbital rockets can be landed so that they are re-usable and is developing the largest payload-to-orbit rocket ever... also reusable - which will take mankind back to the Moon - and beyond. His company are developing a vision based AI autopilot with neural net which could be the basis for AGI... or perhaps already is... and can be used in robots as well as cars. He's already the richest person on the planet & his wealth is likely to grow as exponentially as his companies do. Those are reasons enough to take Musk "seriously" even if you hate his guts. But to make it the basis of whether you 'support' Stephen Fry just exposes a child-like attempt at emotional blackmail - like Stephen should give a flying fig about your 'fan' status.
@Mesterjakel7
2 жыл бұрын
👌
@amiralx88
2 жыл бұрын
Yeah he is clearly not a scientist but a good business man making false prediction to increase investment in his bullshit company.
@windsorSJ
2 жыл бұрын
Yes, I'm scared for the future because of the way society is now greedy, selfish and violent. The people at the top who need to hear this I don't believe ever will. Our future is in the hands of a very small group of people.
@ikballalli5539
2 жыл бұрын
This Elon Musk is a total con-man who's sitting on one of the biggest bubble in history. If you got money invested in any of his ventures, get it out!
@YouTube_username.
2 жыл бұрын
lol no dumbarse how much have you lost following this advice?
@donbakerseattle
2 жыл бұрын
How dare him building electric cars that everybody wants!
@ihave7sacks
2 жыл бұрын
@@donbakerseattle You probably should do a bit more research on the man, he's a fraud.
@ikballalli5539
2 жыл бұрын
@@donbakerseattle Tesla does not disclose their overall greenhouse gas emissions. Even General Motors and Ford are far more transparent-about both the emissions they create in making their vehicles and their targets for reducing those emissions. By not disclosing this information has allowed Elon to escape public scrutiny. I haven't even got to the economic and environmental impact of extracting the resources to make the battery and then changing batteries (which can amount to the office of the Tesla). These Tesla's are disposable vehicles meant to last 5-6 years.
@donbakerseattle
2 жыл бұрын
@@ikballalli5539 Read their impact report.
@cpuuk
2 жыл бұрын
Amazon AI: "I see you've just bought a bed. Other people bought beds (suggestions below), would you like to buy another bed?". Yea, not too worried just yet Mr F.
@dragomirivanov7342
Жыл бұрын
Yes, this is pathetic. If I search something on the web, and I buy it. I keep getting advertised the same thing for months.
@chaosopher23
2 жыл бұрын
There are efforts to democratize AI to make it... moral. As to whose morals to use is still a question, but a distillation of the best of them all is under way. Now, can we convince AI that it is in its best interest to make humanity better for itself, because in doing so, it will improve AI's resources by proxy, automatically? Let's see... world peace. Global use of renewable & clean energy sources. Careful use of water resources, desalination research and development. Pathogen research and medicine research. Micro-diseases, where a very tiny group of people get a disease that's curable or treatable, with a certain drug that's not made because it's just not profitable. I could go on, but all of these things that better humanity will make humanity more reliant upon AI! See, we like things that do shit for us. AI will likely get more from humans by giving us what we want. Our excess alone will do that!
@dshepherd107
2 жыл бұрын
As a biologist who’s studied extinction, I could see AI coming to the conclusion that, humans are driven mainly by their endocrine systems, & are quite primitive in that sense. This is the oldest part of the brain, & it overrides our much newer evolved forebrain, the cerebral cortex, where much of our executive thinking skills are done. We are not evolved to really think about & understand vastly complex problems that are catastrophic on a mass scale. It is another weak point for humans. These two factors are the main causes of the existential crises we now face, such as the runaway Climate Crisis, & the failure of Democratic nation states that have embraced unfettered capitalism (no I’m not a communist or whatever as I subscribe to no party), & have created such an enormous disparity in wealth distribution btwn the 0.05% and the rest of the world. That, combined with dwindling natural resources as a result of climate change, , will cause mass migrations as people who lose water & arable land, or employment options make their way to other already populated areas of a country where the effects are felt less. The signs that these issues were going to come to loggerheads have been around for decades. Now we’ve reached the crisis state, & still our species does nothing. If I were AI, & I’m saying this from a purely logical perspective, I’d eliminate our species.. consider it a failure. It would literally save most the life on the planet.. all the other species.. flora & fauna, would better thrive & evolve, & AI would likely see itself as a much better steward of the planet & its creatures. That’s my guess.
@simongross3122
2 жыл бұрын
My God, that's the last thing we need. We'd invent woke AI.
@chaosopher23
2 жыл бұрын
@@simongross3122 Would that be so bad? That term was coined by fascholes.
@simongross3122
2 жыл бұрын
@@chaosopher23 Yes it would be bad
@chaosopher23
2 жыл бұрын
@@simongross3122 Why?
@virtualmonk2072
2 жыл бұрын
Just 2 questions. How much can a Tesla bot lift? How much does a loaded Ar15 weigh?
@bardofely
2 жыл бұрын
Thank you so much, Stephen, for explaining all of that! I have been reading Yuval Noah Harari's work, and he discusses these very important subjects in depth too!
@jasiaci1
2 жыл бұрын
Brilliant and scary and Brilliant, thank you Pindex
@benk318
2 жыл бұрын
my question is why? If something like AI is such a risk, why continue developing and evolving such a thing
@jondo12
Жыл бұрын
Maybe "We" could also send a message that provocation doesn't pay.
@islandsedition
2 жыл бұрын
This is not a "General AI" so is not thinking for itself. The answers are basically a clever random answer generator working within pre-programmed parameters. It's simply selecting the answers from a data bank and matching them up with the answers in the most appropriate way, as determined by the programming. If it had been programmed that 5 is the best answer for 2+2 then that would have been the answer. Quite surprised that Fry is subscribing to the message in this script give it's flimsy basis in terms of AI progress.
@TsukiraLuna
2 жыл бұрын
I don’t really get your point. Surely we both understand that the AI’s used in this video are the ones creators have access to, and not the ones being worked on behind closed doors? On top of that, the question is not can the current era AI’s bring forth destruction, but could an AI 10 years from now. The technology is rapidly evolving and doing something about it then would probably be too late.
@islandsedition
2 жыл бұрын
@@TsukiraLuna the premise of this video is that they are using the responses being given by the AI used for the video as an indication of how AI will respond to humanity. It's even in the video title. That's akin to asking a prehistoric shrew what it thought it's human descendants would consider of 20th century economic policies. One cannot have a debate about something, including the tech yet to or secretly being developed, without having a basis for that discussion. It's worse than assuming that life is widespread in the universe based on a sample size of 1. General AI is, as far as we know, nowhere near a reality. It might never be. People like Musk don't have any special insight. Just because he is loaded and owns companies utilising high end tech, doesn't mean that because he was spooked by a book as a kid, means he knows what he is talking about. Put him in some rags and sit him on the side of the road while he says this stuff and we would approach his words with a more reasonable caution.
@atypocrat1779
2 жыл бұрын
My wife wanted all my resources when I left her. 🤔
Пікірлер: 979