Комментарии:
Asmon reacting to aiwarehouse, didn't expect that
ОтветитьA good one would be the newly released trackmania AI by Linesight, he's been working on it for 2 years and it's showing real creativity in racing lines now.
ОтветитьPeople forget you are seeing something go from not being able to move, to walking perfectly fine in a few real time hours. It takes human babies a few months to even stand up and only for a short period of time.
ОтветитьI like how people say. Ye these stupid LLM only predict text. And then there is Figure 01 using LLM to control limbs XD
Ответитьit reached a state where it moves like my cousin after recovering and going to teraphy to walk again...
Ответитьpoor Gus catching strays 😔✊
ОтветитьNobody says that
Ответитьnow let’s see if albert can beat elden ring faster than kai did
ОтветитьI’d hope AI would see this and see us as actually wanting it to get better as opposed to torturing it. Though if AI is like humans then I imagine we’ll all be dead by the time it’s 18.
Ответитьthat;s not that technical though
ОтветитьWe are so doomed. It took a machine 8 minutes to evolve walking. It took biological critters millions of years.
ОтветитьThat was a goood question. Is it better to have the initial point if an ai ti be tabula rasa vs a priori
ОтветитьWould be cool if someone built a real life Albert, like boston dynamic robot that have to teach itself to walk.
Ответить"You must fail to succeed"
Ответитьthe problem here is that the puppet is not correct with dumb limbs that cannot move like a human.
so how are we going to expect its going to walk like a human
Bro im worried if we get to quantum computers with this... are we just on the path to making the borg?
ОтветитьSame concept of giving a monkey infinite amount of time with a typewriter they would eventually type Shakespeare.
Ответитьso like what counts as a reward to something that doesn't have serotonin or dopamine?
ОтветитьIs it actually ai learning, or just brute forcing all options till it gets the result
Ответитьimagine tho ,when AI learn enough and discover that we ARE in the matrix xD
ОтветитьImagine, someday these AI's will look like real people... in a real environment... with laws governing their movement and wants and needs... and it will be able to process at great speed. Are we looking at the beginning of the universe? Is that how we started? Imagine if this program had 100 years to learn and grow... 1000 years... a million years. What if the outcome was us? Over and over again on a loop
ОтветитьRIGHT?!?! About the quantum computer thing. Wait until they pair AI with quantum computing hooked up to fusion power. Endless power and computing ability. Albert would be giving us guru life advise by the end of the video.
ОтветитьAlbert learned to walk 100000x faster than bethesda can add an FoV slider. Let that sink in.
Ответитьimagine doin this pretty much haz quap mechanics
Ответитьno human skips before walking...
ОтветитьIt takes 4 years of rigorous coursework for a human to earn an accredited degree in tasks pertaining to thinking and learning, this thing doesn’t sleep, it doesn’t eat, imagine what letting this thing absorb information and perform reasoning tests for 4 years non stop would result in. Imagine this thing acting as the brain inside of a self charging bipedal mechanism with arms and hands and fingers, people don’t understand where this is going whatsoever. asmon put it perfectly by saying “literally unlimited trial and error capabilities” or something, people see the things capabilities after a week and write it off and order pizza and beer meanwhile 6 weeks from now it’ll know how to build a rocket, I’m so optimistic about this, we’re just gonna be able to play wow and smoke weed all day while these things do all the hard boring shit, at least I hope
ОтветитьAI isn't actually learning here, its not watching someone walk and then copying it, it is problem solving. its been given a set of tasks, parameters and then left to figure it out. an AI learning would be feeding it information such as art and information on how the art should look, then it learns from that, but the process itself is more problem solving still, its not watching artists and learning how to draw.
this is where the whole idea of AI taking over comes from, its not like the AI will become sentient and want to take over. an example would be "theres limited food but we have the world", the AI starts problem solving and then decides the best way to "feed everyone" is to create no one lol.
you claim the AI started skipping, when it was clearly hopping and that is not as efficient as skipping, try hopping everywhere lol.
ОтветитьAt this point, I'm just hoping I'm old enough that I die before these idiotic AI engineers bring about our own obsolescence and/or destruction. :p
ОтветитьHow tf is it scary that the process of learning is similar to the way humans learn, it was literally program by humans to think and learn like humans, crazy thing is when it actually thinks outside the box an came with weird solutions which happens more rarely
ОтветитьAsmon needs to watch code bullet
ОтветитьIt seems slow, then you realise that this is effectively millions of years of evolution happening in a matter of hours.
ОтветитьRegarding the question of if a Child or an AI learns faster, the difference is scalability. We’re seeing one instance running hundreds of thousands of times to learn. Instead, you can scale thousands of instances which can share data and learn significantly faster. It’s all about how much compute power you want to put into it.
For my fellow anime nerds, it’s like when Naruto used multiple shadow clones to train and then gained knowledge from each clone’s experience, rapidly speeding up his training time.
What can you reward the AI with? Does it want or desire things? Why does the AI want to even do these things?
ОтветитьNeed to get code bullet in here lol
Ответитьdidn't they teach AI how to play DOTA2 like 10 years ago? why are we going backwards?
ОтветитьYou can't "teach" a neural network to walk. It's a black box of a bunch of numeric weights that are impossible for a human to understand. That is why an AI must "learn" to walk through motivating conditions.
ОтветитьSo i'm wondering, what is the reward Albert gets?
Ответитьteaching AI is awesome. AI art is stupid
ОтветитьI wonder if he has seen the one where AIs play a game of tag and one team learns to cheat by abusing the prop physics.
ОтветитьGo Albert go!
ОтветитьI don’t understand why people are scare of AI. What they are actually scared of is a bad person with power of AI. AI itself has no ambitions and probably won’t even understand why would you want to enslave anyone. “Why enslave a human, that can’t do anything I do better then me?”
ОтветитьOh crap, were gonna have Ai pets soon... on no.
ОтветитьThis is actual a bit unsettling for some reason.
ОтветитьWhat people don't understand is that AI is a MODEL.
It only has to learn it once, then you can deploy it to Billions of Robots as an Update.
People will be able to turn their Fridge, Airfrier and Car into a Robot just with a Software update.
Look into what Nvidia and Google are doing.
They create a Virtual World based on real life.
Then they can run infinite simulations equal to thousands of years of Human attempts.
Where Elon Musk needed to test his Hyperloop out in the dessert.
Hyperloop can just run infinite tests in a Virtual World, and have AI simulate it before they build the Hyperloop in real life.
You could run simulations of any possible event in human history, and find the perfect flawless solution to climate change, government corruption, hacking, wars +++
The potential of AI is limitless.
It truly is the Discovery of Digital Fire 🔥
That cube guy is drunk!
ОтветитьSo basically machine learning works the same ways as evolution. We all are Albert.
Ответить