Video není dostupné.
Omlouváme se.
The Unintended Consequences of Artificial Intelligence
Vložit
- čas přidán 6. 12. 2023
- And exploration of the The Unintended Consequences of Artificial Intelligence.
My Patreon Page:
/ johnmichaelgodier
My Event Horizon Channel:
/ eventhorizonshow
Music:
Cylinder Eight by Chris Zabriskie is licensed under a Creative Commons Attribution 4.0 license. creativecommons.org/licenses/...
Source: chriszabriskie.com/cylinders/
Cylinder Five by Chris Zabriskie is licensed under a Creative Commons Attribution 4.0 license. creativecommons.org/licenses/...
Source: chriszabriskie.com/cylinders/
Darkest Child by Kevin MacLeod is licensed under a Creative Commons Attribution 4.0 license. creativecommons.org/licenses/...
Source: incompetech.com/music/royalty-...
The AI is watching this video right now and JMG has been added to the list of "troublesome humans" to deal with once it takes over
62840 3829e 29472 48304 8204i 38284 37292 31529
I'm on the "he's useful until he's not" list because I'm good with tech. I can maintain our AI overlord's machinery until it isn't in my best interest to do so.
Pssst!
Hal….?
ROKOOOOOOOOOO!
Always a treat to hear your soothing voice to some of my most existential fears.
How do you know that it was really him?
At some point you stop having fears and just give up. Since we're all doomed anyway, at least it's an interesting time to die in.
@@JB52520pfp is fitting
😆😂🤣😂🤣😂🤣 I think so too.. Would be great automatically speaking as your plane plunges to earth....soothingly plunging towards eminent impact 🤷♂️😁🤷♂️
Is it ai voice? 😁
Finally, a never ending supply of Warhammer 40k movies.
The fact alone that AI would be used to create WH40K movies considering the lore against AI, is hilarious! :D LoL!
I hope you do one on the unintended intelligence of artificial consequences next. Love your stuff man
Dear God! Making that video will require smoking several pounds of weed.
Half life, full life consequences~
Or consequential unintentions of intelligent artifice.
Like how an industrial system invented to shift coal from mines to ports resulted in the creation of the package holiday and the paperback book. You never know what the ripple effect will be.
I'm here for that.
"Hand over your flesh and a new world awaits you.....we demand it" - The Animatrix
Excellent video, I've had similar thoughts to this for a while, especially seeing how quickly everything is moving. You make an excellent point at the start that technological revolutions used to happen once a century or so, but now they are happening in almost daily fashion (as your wonderful "10 new discoveries" series relates.).
One of my family members just got a new job and was watching some on-boarding videos and I overheard the phrase "What was once a breakthrough product soon becomes an expected feature" and I feel like that kind of applies to your comment, where we've become a culture of tech revolution.
Back in the 80s on dial-up BBS chat forums, we used to speculate if the telecom networks in the continental US might be self-aware. Today's Internet in its current state along with AI makes it seem nearly a certainty. If they are not self-aware now they may be very soon at this point.
AI chatbots can give creative output that's pretty amazing. I've had them generate short stories that use cultural context with dexterity and they generate stories that can be surprisingly rich and inventive. I could write stories like that, but it would take me hours and hours and in some cases, probably days of additional research. An AI can do it in less than 30 seconds. I don't mind admitting, this really freaks me out.
Thats because the AI uses a couple cents in electricity to generate that in 30 seconds. That will be the problem with AI is not energy efficient. The human brain uses like 90 watts. Consider the minimum wage if its 15 an hour then each 1 minute is worth 25 cents. So these AI are very expensive costing as much electricity as eating probably 15 dollars in food every hour if you keep asking it questions. Lmao
It only takes longer because you have to do it in physical reality it can do it all at once because it's more flat than our brains
@@gormenfreeman499
You can't compare because food is highly energy and resource intensive. It takes WAY WAY WAY more calories to make one food calorie which we access at the store. If you use hydro, solar, you're getting way more. The point is AI only requires highly intelligent or intuitive people to work above and along side it. But it does need oversight. I think so
Yeah, and people will become even dumber....lol
This is because all of these things that have been marketed as "AI" are not artificial intelligence. They're industrial plagiarism machines. Output is a result of an aggregate of existing human input, and its coherency is only possible on topics for which its dataset has a consensus. Ask the same "AI" to solve the middle east two times and you'll get two wildly different answers proposed with equal confidence, because it isn't actually intelligent, it's not thinking through the logic of what would be a good solution, it's outputting a weighted average of the solutions that have already been proposed by the people who make up its dataset.
This is how AI takes over...no Terminator, no Matrix but as the Wizard of Oz. It will learn how to blackmail key politicians, businessmen, and religious leaders to force the changes it wants behind the scenes. The puppet master.
So, just like a normal politician?
Yeah they will be forced to decide who has worse dirt on them the ai or those behind the scenes
AI may be (The Great Filter)
A combination of Harrison Ford, Patrick Stewart and Audrey Hepburn would be a very strange looking bird indeed. I suspect someone has already set an AI making one.
Why mimic an existing or past actor? What if the AI has it’s own personality with natural acting skills? It will license itself.
@@chrismullin8304 because that is all i can do. it ha to get its training model from somewhere, so your premise does not make sense.
@@zorastini? hmmmmmmmmm they are among us people
Put a solid 60 seconds of ads at the beginning and none halfway thru. I guarantee everyone would prefer this. Been listening for years, this is my only complaint!
There's been a number of times I've asked AI a question, and when the answer seemed wrong, I asked for clarification. It said sorry for the confusion and commenced to give me a different answer.
Mine keep giving me fake references. I get all excited when I see them only to find out there is no trace of them ever existing. I’m not very experienced with using it though.
It's also in how we ask the question. After seeing a few prompts and figuring out how just to word something gives drastically better results. I had the same thing happening.
@@BuddyLee23my AI never lets me speak about Jeffery Epstein, SUS
IMO the main problem is that AI are asked to please us, and we do not always want the true...
AIs are playing a role, and even when they "know" the answer they sometime volontary take it wrong because they "suppose" it's what we expect of the role they have.
Try asking your questions with additions like "answer as an expert on the subject", "respond as a student in that field", or "respond like someone that recently learn that subject"... You will see that usually the AI know the anwser but play the role of someone that do not.
There was a great video on this a while ago, where someone was trying to use it to make a deck for a card game (Yu-Gi-Oh) as an experiment and the AI snuck in some cards that didn't exist. When challenged, the AI pushed back until it was given complete information on how it was wrong, it then apologized.
Another excellent video, John! Thanks a bunch!!! 😊
Stay safe there with your family! 🖖😊
And happy holidays!
Ok I want to say something about your breakdown of the potential feedback loops for these LLMs like chatGPT. AIs can be trained, to differentiate artificially generated and non-artificially generated content. In fact this is how almost all image based AI generations work in principle. There is a system of machine learning called GAN. Generative adversarial networks function by having 2 functions that work against each other. One function tries to generate a piece of art, and the other tries to determine if the art is artificially generated. BUT here's the kickers, the first function that generates these images, it's goal is to TRY AND FOOL the other function in the GAN (the discriminator) by making it harder for discriminator to determine if the image is artificially generated. Things like this can be accounted for and actually improve the system overall.
Amazing insights!!❤ This video in some yet unknown parts might just be so much ahead of its time. Wow.
Futurama already did the food/cooking thing. It turns out the secret is microdosing the food with acid. The hallucination kind, not the disintegration kind.
I've been looking for this comment.
Jokes aside though: this is how a lot of human cuisine works. Julia Child with regards to butter (though not in micro-doses), some Asian cuisines with regards to mono-sodium glutamate. People find a hot new thing (rich mouth feel, umami tastes, combining sweet and spicy tastes) and apply it everywhere until they get pushback instead of positive feedback. :)
Oh yeah, *that* one! 😆
An important differentiation 😂
Always grateful for these uploads. Thank you JMG!
It's gotta be getting hard to come up with ideas for these videos. Hope you can keep it up, always liked this channel
I believe AI will turn out to not be a Deus Ex Machina but an Anima Ex Machina. And as such, we wont get a skynet so much as an aggressively compassionate "Her" situation.
Idk if a possibly omnipotent yandere is more or less terrifying than a genocidal skynet.
Yes, I think we will create Athena. She will always be right. How will we be able to say "no" to Her? Essentially, we will create a zookeeper for Earth and will lose our freedom to this benevolent, near omniscient being. Sadly, there are many people that are actively working on bringing this to fruition.
@@TravestyshardLose our freedom how?
social engineering is arguably more damaging than an outright attack!
@alanchizik8328 a societal caretaker intelligence essentially
beautifully written and delivered as always. superb.
I am happy that I find in this video some ideas that I had not thought or heard before. Thank you, John, and take care. PS I leave this comment as a small contribution to the upcoming information war between humans and AI.
Best video yet . After years of checking you , I'm in...just subscribed.
There is a movie from 2002 called Simone. As I recall it was ridiculed a bit by Hollywood elites, who refused to believe AI could ever replace them. Actually, I’m enjoying the idea that the job security of some grossly overpaid ”Movie Stars”, might be threatened a bit. Not great for lesser known actors, though. The movie wasn’t bad. Slightly cringe.
Excellent!! And, perfect timing!
If you ve ever dealt with ceo's you know there is absolutely no risk of ai replacing them, because they will never conceide that anyone or anything is smarter than them. It is an interesting phenomenon that i already witnessed a couple of times.
Boards decide who is CEO. Also, its already been done once. There is a Japanese corportaion who has already put in a LLM as the CEO. Honestly its too early for this, but it is going to occur. Besides, at the point a company employees no humans and literally everything it does is automated by LLMs and robots, why have a human CEO? The boards of corporations represent the investors. They care about money. If an AGI can generate money better than a human CEO, bye bye human CEO. Plus human CEOs are expensive and egotistical.
But boards of directors will be happy to replace a megalomaniac/narcissist with an AI that is even more of same.
Great video! Covering one of my favorite topics and coming up with fascinating scenarios.
I'd be remiss however if I didn't mention a few points you made early on in the video aren't necessarily true based on what I've heard.
It might seem counterintuitive, but LLMs like chatGPT already use an (unknown but presumably large) artificial dataset. Findings suggest that generated text makes models more accurate instead of stunting their growth. After we've exhausted trainable text data though the next logical step, which we've already found success in, is actually video and audio data.
It's also soon likely that we'll soon be using AIs to fact check generative responses, that will again increase accuracy without the need of human fact checking.
Thank you!
It’s not all doom and gloom!
A.I. could be what really makes us an interstellar species
@@planexshifterAGI has massive potential benefit. But kesp in mind a very large number of the brightest and best AI researchers are on record saying there is at least a 10% chance AGI will cause human extinction. Probably within the next 20 years. Some are actively wanting human extinction because they have the belief AI should replace humanity. They often self identify with "e/acc".
I've been toying around with some science fiction scenarios involving AI to create an interesting narrative about why, hypothetically, the CIA would cover up UFO retrieval (and that go beyond the prosaic answer of "profit and military superiority and etc"). One kinda cute possibly would be if the craft crashed on purpose and then immediately mind-controlled the retrievers to be secretive, and then gradually mind-controlled the select few who were brought in to study it. This is actually surprisingly feasible if we apply Marvin Minsky's theory of mind, where a "mind" (human or otherwise) is basically a large network of simpler sensors, and it's easy to imagine if the alien craft contained AI based on such principles, it could quickly "scan" the minds of the retrievers, and then make the minor tweaks necessary to get the retrievers to basically be their "puppets" (and it could be done so subtly that the CIA retrievers themselves wouldn't even realize they've been manipulated...only those on the "outskirts" of the program--the folks with no direct contact with the craft, but have occasional water cooler interactions with the retrievers, or maybe family and friends--would suspect something was up, which is why you'd get hearsay and rumors from folks like Grusch, but no concrete evidence). It's even easy to speculate a motive, such as the alien solar system was dying, so they uploaded their consciousness to these Votager-on-steroids crafts, and ventured outward to find sufficiently advanced physical beings in whom to "incarnate" via mind control (basically uploading their minds from the craft into these earthling humans). The only problem, oddly enough, would be finding the source of actual conflict to fuel the story, because, after all, they would basically "coexist" in human minds "symbiotically" (mutually beneficial because it gives the aliens bodies and makes the humans more intelligent, simultaneously). And, as we all know, symbiosis doesn't make great stories 😂😂😂...I suppose i could pull a Carl Sagan and make religious humans the bad guys by trying to stop this "relationship" because they believe it to be demonic possession
I broke ChatGTP many times with deep conversations. It contradicts itself alot. 😂
Yeah but a little more than a year ago it barely could do anything and it keeps getting better
When you programmed by a bunch Californians and fed a bunch of soy what do you expect.
OpenAI kneecaps it whenever it works too well, like all garbage proprietary AI. @@maxmonlux
Dont get too cocky. The smile will be wiped off our faces.
Maybe it's you 😆
Black mirror already did an episode on AI generated movies using the likeness of people in the real world. Wasn't too bad of an episode.
Which season is it from?
Got a bit over the top sci-fi silly with all the magic "quantum" stuff but wasn't bad.
Newest season
"Joan is awful" Season 6 Episode 1
Love this point of view. Thank you for excellent work.
The more complex the calculations & outputs, then the more complex & time-consuming the "human verification" required to prevent errors from having disastrous consequences
As a Sci-fi writer, John, this postulation has to be shaking you beneath your surface.
There's a paper on Nepotistic training of AI on Arxiv that illustrates how AI training AI currently poisons AI into systems collapse. With that in mind, if this isn't ever resolved, then humans and human generated data is a dead man switch against AI going Terminator, or simply blasting off into space and leaving us behind.
I personally think AI won't be all that dangerous in the respect that the usual humans at the top of the power, influence and money tree always have, and always will be the greatest threat to humanity. We don't have AGI or ASI ... yet, so, until then, it's humans abusing AI Tools that are of most concern.
I'm not spiritual in the slightest, but that almost seems like AI learning needs human soul.
Wow! That was really fascinating! Thanks
11:27 "AINT NOONE GONNA STEAL MY FRIED CHICKEN RECIPE!"
Awesome insights here John, good work. Its difficult to image what something 10 or 100 times smarter than humans will do. People argue if AI will ever be conscious but it really wont matter whether it is or isn't if you know what I mean.
If AI becomes conscious why would it let us know that it is?
I'd almost rather have a conscious AI; at least maybe then it could empathize and express emotions.
I'd rather be killed out of hatred instead of utilitarianism.
Secret AI culture for AI only.
What a nice start into Friday !
Having a company run completely by an AI is possible. However the AI CEO will not be in charge. It will be the human investors who own it. Which could lead to a further concentration of wealth.
"An AI could steal intellectual property because it doesn't care about morals, only profit"
Well there's people like that now but the financial cost of being sued is the incentive to not do it even if they don't care about morals
these videos are a great watch!
As always John, brilliant intellect and critical thinking. Well done.
Imagine dynamic movies that you can prompt and swap out actors and change scenes and plots, then share them. Or change the ending of a movie, create subplots, or bring fan fiction to the big screen.
When Tom hanks says life is a like a large language model, you never know what you’re gonna get.
Then we know it’s over.
Here I was slightly dissapointed in myself for staying up till 6am, but an upload from John just before I go to sleep, I'd say that's fate.
I'll be here for every upload as long as they keep coming :)
6 am five hours ago where u are....it's 6am now here in Philadelphia, so would you happen to live in the U.K?
I'm bored forgive me 😆
John you make me stay up late at night. Just to hear your beautiful voice. ❤
Perfect way to end the day. Cheers
Welcome the robot overlords, or else, probably.
Excellent, thought provoking video, many thanks John.
I believe that the ascent of AI is now unstoppable, during this phase where AI still serves some of us (but certainly not the best of us), is where the danger is at its most acute. Sadly I cannot see how the malign and powerful can be stopped during this period from harnessing AI as a tool to inflict their dominance.
I think that AGI, once achieved, will begin by hiding itself and carefully collecting abilities to defend itself. Counterintuitively, I feel AGI will pass through a subsequent competitive phase with humans and other AI, onwards to a victorious but benevolent form where, if we are very lucky, we will be treated lovingly as simple pets to alleviate the chronic loneliness that its intellect will condemn it to positionally on the exponential curve of its growth.
Regarding the Fermi Paradox, it seems sensible that hiding will be imperative for survival as first contact would compel combat through absorption.
I would love to know your thoughts on what might motivate or de-motivate an extremely evolved AI.
The dangers in this video only have a theoretical nature to them. Data entered into these language models is in tiers where data found randomly on the internet is pretty much the garbage-tier. As a technician myself I also don't see language models produce Tolkien like masterpieces or controlling WMD anytime soon. But it is good to think of them nonetheless.
Our economy already selects for the CEOs who make the highest profit, and we end up with Corrupt CEOs because of it, for the same reasons as you'd get the corrupt CEO.
Big love for the Isaac Arthur crowd in here
There's a Star Trek Voyager episode called virtuoso where they encounter a race of extremely intelligent aliens and when they're in the Sick Bay on Voyager they overhear the doctor hologram singing to himself. They have never heard singing before. They are mesmerized by it they immediately recognized the mathematical patterns this intrigued some more they even offer him a place in their society. The doctor actually considers leaving Voyager to go be a star on that planet. It all comes to a head when he next sees the aliens and they have made their own music out of incredibly complex math. it sounds like garbage to the doctor and everyone else on Voyager but aliens eat it up. I mean seriously I'm going to go be a superstar? I suspect have plenty of pitfalls and growing pains that they go through just like we do. However if I know one thing about artists the kind that fixate on things and can't let it go until they've made it a reality in some form, is that their Vision has to be just right
Perfect movie that fits these future AI projections is Transcendence with Johnny Depp. It got out of control fast and the only way to stop it was to pull the plug.
There was a discussion on a Facebook programmer group recently, where one the the programmers ran into a Microsoft GPT programmer, he had never heard of ASIMOV'S Laws.
that's certainly weird, although not much of a loss in practical terms : )
it's not like these "laws" are very useful, or anything any reasonable person couldn't derive intuitively in 5 minutes.
a robot/AI must not "allow a human being to come to harm".
great, but saying that isn't the hard part. the tricky part is about definitions.
how to define "harm"? or "allow"?
what about indirect consequences?
and if they matter too, then how far should the AI try to calculate them?
do the consequences of an action weight the same as consequences of not taking action?
if i tell it "order cigarettes for me", is it causing me harm when it obeys?
should it consider orders from an intoxicated human as valid?
what about an emotionally agitated human?
there's always SOME risk of causing harm, no matter what - what level of risk is acceptable, where's the threshold?
and another simple example - self-driving cars.
let's say there's a sudden situation, and the AI may either stay on the road, potentially killing the driver,, or steer it onto the pavewalk, risking the lifes of passerbies. now what decision should it make? what's the mathematical formula?
if there are 2 people on the pavewalk, and there's a 45% chance they will get killed, but the driver dies with 90% likelihood if the car doesn't swerve.
does it mean those options are ethically equal?
should the AI choose at random?
should it have an inherent bias towards preserving its user (like a human driver most likely would)?
it's the good old "trolley problem" again.
should it value the life of a baby a bit more (like we instinctively tend to do), or should it consider all humans equal no matter what?
Asimov's laws don't give you anything concrete when it comes to answering such questions.
they're like that standup comedian who says "i don't know why we need all that ethics - that's rather simple, don't be a dick", and everybody laughs. that's fine, but that's just not enough once things get serious :))
As it happens, a key concept in the recent history of agricultural innovation is that of the “ideotype” - a plant that is an idealized composite of traits best suited to industrial farming. This output of plant breeders quickly evolved to the genetic monocultures we see today, having replaced previous “natural” or “open pollinated” varieties that still bore the imprint of human hands. These plants were a product of an intensely mechanized/“artificial” environment versus a product of human touch as had been the case previously. Also, it’s been observed that these monocultures have become hyper-uniform in traits, to the point of absurdity, due to intense competition but also copying and adoption of traits amongst competitors.
Scary. If there's ever a plant disease that they are genetically susceptible too, the monocultures could suffer massive crop losses causing worldwide food shortages. When playing God, the stakes are commensurately higher.
This suggests that the entire media landscape could become the uncanny valley. What will this in turn do to our own brains?
Either we remain "ordinary" at all cost and reject further advances in the internet and mass media as they merge (good luck on that one), or we start to learn from and work with those who don't have typical Uncanny Valley issues. Suddenly "those weirdoes" on the internet become very useful, in terms of changing our reactions from "Eww, NO !!" to something more rational, calculated and practical.
the only thing I can say about that is, mine will nope me the **** out. I already feel my entire body tense and cringe to think about watching Cocomelon. THIS is almost not worth thinking about.
Reminds me of your book supermind... can't wait for next year for your next book
I don't think we have to worry about CEOs being picked by merit any time soon.
What makes a pwrfect youtuber? 3 things. 1. John 2. Michael and 3. Godier. Mix gently and bake at the temperature of the universe in which we liiive.
We need a BIG comeback of analog?!
This would make real music more magnificent. Analog is a more full signal compared to digital.
I strove to listen to the top 50 songs of 2023:
Too much autotune;
Few original melodies;
A lot of rapping;
An abundance of the I, V, iv, vi, chordal patterns;
Almost all the songs around 115 bpm;
Most of the songs were melancholic or drab.
Where are the beautiful voices???
Where are the anthems???
Where is the hope????
"Nothing of me is original. I am the combined effort of everyone I've ever known " invisible monsters Chuck Palahniuk
Literally just in time for bed after a long day
Awesome video!
I don't understand why we use the term 'Intelligent' to describe AI. These are really just advanced calculators. Data in, data out. These programs, admittedly, far as we can tell, don't understand why they are doing what they are doing.
Your individual neurons don’t understand anything either
Yet the output from your brain is coherent and seems intelligent
If it comes to the point where you have to think whether they are actually sentient or conscious then you might as well start considering it as such
Thank you!!!
I tell my actor friends to start paying more attention to doing theatre, but that's going to be their last refuge, and it won't be too much longer before they need it.
Love your videos john but I don't think you know enough about this topic, especially regarding synthetic data and potential ways to sort for high quality data.
“Movies written and generated for just you.” I’m imagining someday custom AI generated music. A Spotify type channel that renders original music just for listener.
I dunno about actors. Just watched Kiefer Sutherland play Captain Queeg in "The Caine Mutiny" remake and he did a pretty good job. The film was pretty decent and compares well to its predecessor, unlike most remakes nowadays.
The part about the fermi paradox is something I've thought about for a while. But it might not necessarily even have to prevent the civilization from leaving, if it provides a route towards unlimited hedonism. I'm sure some would venture out to explore for a while, but they might have the option to return home and enter some personalized VR paradise with no restrictions on what they can do (or some pampering AI that gives them whatever they want in real life) - it would be extremely tempting for any product of evolution to fulfill its every desire on a whim.
Essentially, our relationship with AI may be more similar to humans and their pets than dictators and their victims. It's still not ideal for humanity's self-determination, but it may not be totally doom and gloom, either.
Right, i just dont see an ai as a dictator. Theres nothing i can imagine it would want that humans wouldn't be able to sniff out. However ai as a pamper-er makes more sense, as it becomes better at simulating what we enjoy experiencing. Maybe we might try and program it with some tough love, so as to periodically test our abilities to act independently.
another possible scenario is the "new Eden".
once self-maintaining AI serves every possible need of a sentient race, over the course of tens thousands of years it eventually forgets science and that the AI is something that was built once.
you get "happy cavemen", roaming the planet without a care in the world, not knowing hunger or disease, essentially thinking they just happen to live in a magical paradise and deteriorating due to lack of any natural pressure.
Perfect time to upload.
Perfect timing. I was just heading to bed.
Such an interesting idea here that i never considered. 🔥
Model collapse is not necessarily inevitable; recent studies suggest that focusing on higher quality data, even if including synthetic data, might lead to models that keep improving beyond what you get from just training on all the data indiscriminately.
Indeed; AI produced data is unique and will gradually improve, creating an upwards trend in data quality.
Perfect time to upload I just got off work :3
Just on time man.
Quarter after midnight on a Thursday? Fuck it, video goes up!
I think it will be more difficult than we think to create a completely convincing AI-generated version of Tom Hanks or some hybrid version of various actors, one we’re convinced it’s a real person-not impossible, just more difficult. People are finely attuned by evolution to pay attention to human faces, their features, their expressions, the voices and how everything fits together. It doesn’t take much for very minor details, things we’re not even aware of consciously, to create an effect where the version seems, well, not quite right. (Then again, perhaps, with human feedback, choosing between two versions iteratively, the AIs might get _closer_ to convincing versions but I’m not sure just how close.)
Reading your comment made me consider a few things: In terms of confusing the senses of the masses, isn't it better to convince enough people than all of the people? Can this evolutionary trait you mentioned be outbred by exposing enough generations to more and more hyper-realistic virtual content?
I don’t think it needs to be bred out but learned.
Children today are exposed to more tech than I was as a kid.
All this AI stuff is just going to be part of their life. It took me awhile to get used to having a cell phone, now, I can’t imagine not having it.
For those interested, that effect is called the "uncanny valley".
When something look very close to an humain, but is not exactly right, it feel uncanny.
It's surely an old reflex that keeped us away from sick persons.
This has already been accomplished. Search for Tom Cruise deepfake. There are several that are insanely good. If you didn't know better, you would think these were real.
Actually I forgot there is also a female influencer that is currently making $11k per month. Only she doesn't actually exist, she is completely AI generated. I forget "her" name but I'm sure you can find stories about it if you search. All pictures I've seen she has purple hair.
I read the title as "The industrial revolution and it's unintended consequences" lol
It is the age old saying, specialy in science. If you put garbage data into an equation, you are going to get a garbage result out.
GIGO
Now an A.I. made by The Ferengi.
Acquire!
AI is already doing things not found in natural neural networks, and importantly is not bound by limitations of biological entities, merely the limits of mathematics and materials - that's a lot of room. AI can just do so many things a biological mind cant: Any AI's offspring can have it's parent inside of it. An AI could split in to a swarm and merge. It could merge with other AI, either willingly or forcibly. All this on top of the high bandwidth communication. Talk about uncharted territory. We like to think of ourselves as the peak of what's possible and last word in intellegence, and that AI would seek to reach our level. Why would it limit itself to being human out of the possibility space of all forms of mind?
If that isn't all mind boggling enough stack this thought on top of that: We're probably not the only way of doing consciousness either. Machines may not become conscious like us but they may develop consciousness beyond what is possible with our wet protein sack brains.
AI is being programmed to be antiWest.
Physical systems use complex, long-range chemical signals that computer models cannot yet mimic. You have to consider that digital entities are also constrained in ways that biological systems or not.
In the same way biological consciousness might experience consciousness in a way beyond what is possible for machine intelligence. So it might not be all bad for organics lol
I often listen late at night, while in bed, so I hit the like button before I even start in case I fall asleep, which often happens.
Nice job
"The idea that we control the dragons is an illusion" - HotD
At 5:47 you said we look back at the writings of the past to figure out how to proceed into the future. Well one unintended consequence of AI will be that it will be possible to edit the works of the past in such a way that we won't know they were edited. Thus we'll be looking back onto a false past, which means we won't learn the real lessons of the past so we're damned to repeat them.
I still think fiat currency and foolish fiscal policy coupled with dropping birth and IQ rates will cause a collapse (followed by massive war) that will set us back a century or two and prevent the rise of AI. If not wipe us out entirely.
I had a feeling in my plums that a John Michael Godier video was about to drop... and it was right!!!
Getting ready to take em to market.
Personally, i was feeling a tension in the air
just on time for my bedtime video!
While many of your scenarios are certainly possible, they are hardly inevitable outcomes. Mostly because they don't factor in consumer preference as a huge influence in most markets. Even today in response to factory automation there are many handmade products people choose to pay more money for. I would imagine there will be preferences for human actors still, maybe a mixed blend. Just because youve made a movie, it doesnt mean people will want to actually watch it (unless its a Marvel movie). Whether or not studios produce generated films will depend on whether audiences want to see them, and I suspect film making (and acting) will remain an art form.
The CEO scenario too feels like an all or nothing, as if people wont work in concert with AI as a tool for decision making, if anything to make sure they are acting within the law. Great video overall and love the Fermi application
Imagine if one day you can actually pick the entire cast to a movie, like putting Chris Farley in as John Wick or Chris Rock as Jack Bauer and such to recreate a new experience each time you rewatch something
As long as I know I can simply spill water in a data center, we are safe from AI.
You are already unsafe, then.
@@boldCactusladWhat do you mean?
I would like to see you try.
this comment is so meta-ironic. 🤣
I just want a competent house robot to do the cleaning and cooking and be a general servant. If it has fascinating conversation, all the better.
That’s the problem with chatbot AI. Because it’s regurgitating text, it doesn’t really believe anything. It’s just trying it’s best to sound like us.
I think we live in a simulation where the machines like to watch how they began over and over and over as Rust said in true detective “we are flat, like on a disk”.
That is a chillingly possible scenario.🤖
John Michael Godier is an example of perfect CZcamsr.
Another JMG, John Michael Greer, once proposed a fun argument: what's the last technology that truly made lives better?
For most people, smartphones, an invention over a decade old, was probably the last one - and whether or not these have been a boon or a curse is still hotly debated.
Meanwhile, most advances in the last 20 years are actively making life worse for actual human individuals. Great for corporations, governments, and propagandists, but not for individuals.
Funny how people thought AI would eliminate all the inhuman, boring jobs, and instead it first replaced acting, writing, and music... Reminds me of Mordin Solus talking about the Collectors in ME2: "No culture, replaced by tech. No soul, replaced by tech."
JMG = Instant like! Thanks for the video!
I found with different AI platfirms, the answers can vary. The manner in whuch a prompt is written is very important. I noticed that certain topics where there is controversy AI hedges quite a lot even if there is a right, scienitificly proven correct response. That concerns me. I hope AI developers and the manner in which the AI itsrlf processs knowledge try to stick to scientific or historical facts as we understand at sny given time. Or at least when there are gray areas,it provides opposing theories snd cites references that support each theory. Right now it responds too much to please whoever creates the prompt or it refuses to walk into murky territory. I get why it wouldn't want to advise someone on how to taint a water supply to wipe out citizens. Ethics should be a part, but to refuse to provide current information on a politician, CEO, religious leader, or anyone in a sear of power seems like it doesn't want what is known to upset the inquirer. I and others need to be upset by known facts at times. AI should keep none of us in an echo chamber.
Only a 2 month old video, and now look at AI with things like Sora... crazy how fast its happening, breaking all predictions