Talks relentlessly about his AI company with NO DEMO

John Carmack fell off

https://www.youtube.com/watch?v=rQ-An5bhkrs&t=11303s

>>105619826 (OP)
Hello?

>>105619826 (OP)
People have unrealistic expectations of geniuses.
Either you're infallible authority on every subject forever or you're a fraud.

Carmack had his time.
Is he a scientific genius discovering new ways of computing?
No.
Was he an incredibly determined engineering genius capable of picking the right discoveries and implementing them in the most optimal (if not performance wise, then in product from scratch-to release time) way, basically creating the FPS genre and pushing it to its peak and becoming incredibly rich in the process despite his neurodivergency?
Yes.

That's a very good life in my book.

I want to see John Carmack and Jonathan Blow in a cage match.

>>105620142
He should stop the larp then

He does have a demo, you retard. It's the Atari and robot arm thing because he thinks the path to AI is fundamentally a different approach than what most people are doing right now.

>>105619826 (OP)
Computer nerds look like this now ??
Holy

>>105620376
lame ass demo and barely one at that. Hardly mentions it

>>105620417
>he doesn't have a demo
>nevermind he does but I just don't like it because I am a brainlet

kys you retarded faggot
all fields

>>105619826 (OP)
AGI is about running thousands of experiments with different RL agents and environments and learning incrementally and slowly getting better at it. Carmcuck needs more people and compute if he wants to succeed at this.

>>105619826 (OP)
This AI grift trend made me lost a lot of respect for some great people. Carmack. Lecunt. Tao. Just disappointing.

>>105620408
he was always a gym rat, go read the book

>>105620511
It's a figure of speech you fucking hulking neanderthal. Saying someone doesn't have something can mean they don't have a thing that's good. Read a book. My point is valid. You're bald.

He should work on robot hands controlled by meta's emg wristband. He even worked at meta for fuck's sake and he is obsessed with BCIs. It is the perfect fit.

>>105620687
>spends all day in the gym
>whines to zucc that metaverse is impossible (even though vr chat has existed for 70 years)
>leaves when zucc discovers vr chat and asks why meta isn't working yet
>starts (((AI))) company to suck money out of investors
I'm not sure if this guy or geohotz is more of a scammer, but they've got to be close at this point.

>>105620717
>I'm not sure if this guy or geohotz is more of a scammer
why is geohotz a scammer, he gives really shit and dumb opinions on things sure

>>105620376
so his demo is 2010 tech but worse? Awesome genius right here. Reminds me of this math prof who had s field medal believing he could just flip to deep elarning in 2015. He did some.binary circuit net thing and presented to the deep learning departmenrs, showing it could get 90% performance on mnist in a few seconds. Awkward silence in the room because those results were bad by 1995 standards, but he eas convinced the silence was awe at his unparallelesd genius. He never published or presented after that despite spending years pushing his idea

>>105620759
>buys exploit and wins a competition with it
>creates 25k$ (((AI))) box that you can make for 5k$, cries about better, cheaper, faster, more energy efficient alternatives coming out and claims his is better without elaboration
>cries about AMD ROCm instead of just using Vulkan compute like a normal person (likely to try and get more handouts like usual)
there's more I can list, like his self driving car,

>>105620408

>>105619826 (OP)
John Carmack, like the most famous tech bros from early 2000s, was an opportunist. He was at the right time with the right set of skills.
Nothing more, nothing less.

>>105620344
He's not LARPing. He's still trying and still seeking out novel problems and examining them in good faith. The other anon probably had it right, though: his mind is tuned around low-level problems. AI is pure abstraction. It's not about building black boxes from bitwords but from data structures that couldn't give less of a shit about binary logic because AI these days is n-ary in function, where n is not only unknown but sometimes functionally unknowable.

Being able to fit exceptionally complex, low-level logical structures as absolutely a superpower in Carmack's day -- and that's where his genius was. Not only could he do that but he had an insatiable obsession with learning about it. You probably need both because raw talent and genius aren't enough to bridge the gap to accomplishment. It's not only possible but plausible that, as genius and obsessive as Carmack is, it's as far-removed from AI as it is from baseball.

>>105620687
How could anyone possibly have enough free time on their hands and enough interest to read about nerd drama from 30 years ago?

>>105620142
>Is he a scientific genius discovering new ways of computing?
>No.
Well, he did code the best game ever made out of thin air.

>>105620330
Carmack would fucking obliterate him. Have you seen them guns?

>>105620715
Do you put yourself to bed after that much whining? Don't forget to take your ibuprofen if you're going to move goal posts again.

>>105620408
he worked out for years with nogains
then he got old enough where you normally get fat from your metabolism shitting itself but he gets jacked instead. either that or drugs but i doubt he is that vain
pretty big W

>>105621276
no he didn't. he made a mediocre game at the right time.

>>105621315
There is no way in hell he isn't on ""TRT"" at the very least.

Tim Sweeney is unironically the superior Carmack. At least he's going to follow through with the metaverse, while Carmack plays round robin with a bunch of shit without accomplishing anything

at meta he worked in a company with 10bil budget for VR glasses

I guess carmacks work would be more exciting if he was at grok or something, AI seems to require a lot of resources for interesting stuff to happen.

>>105619826 (OP)
I always thought Stephen King was overrated.

>>105620330
All jblow made is a mario clone and drag and drop tablet game. And no, unfinished toylangs like jai don’t count. There is no competition here.

>>105621377
>DOOM is a mediocre game
How completely starved for attention are you?

>>105620910
john carmack peaked in the 90s wtf are you talking about
when doom 3 came out he was known to be a hack

>>105621592
you're a good person for giving her some.

>>105620408
That's a shoop.

>>105620408
would

>>105621639
It's definitely not.

>>105621140
>How could anyone possibly have enough free time on their hands and enough interest to read about nerd drama from 30 years ago?
yet here you are shitposting, its a short read, i finished it in one sitting.

>>105621410
it abaolutely doesn't googlenet was a joke, lenet5 didn't need resources

>>105621592
You never played doom.

>>105621712
what interesting AI tech people use today doesnt need 10mil worth of servers to crunch data?

>>105621719
You need to find healthier ways to get attention.

>105621377
Guaranteed (you)s

>>105621662
Fuckin' A.

>>105621113
Its been decades since he made anything that anyone considers leading edge, even the stuff he did with Aerospace and VR was pretty inconsequential in the grand scheme of things. Compare that to Doom which invented a genre and, more importantly imo, Quake which still had functions that still do the vector math for UE5 and others.

Also, what he presented is not low-level at all, nothing in his wheelhouse in software. It was robotics and mostly applied classic RL to covnets. He admitted to being too low-level to start; writing C++ writing CUDA kernels directly, then using cuBLAS, then cuDNN, then finally moving to PyTorch just so he would have a working demo to experiment with. While having Carmack write a super efficient CUDA kernel sounds like something he can do and be relevant, its not what's happening. He's simply retreading ground OpenAI and Deepmind did in 2013 - while trying to see if he could do something different, so far he has not found anything novel.

>>105621726
theres chat bots that run on KB of memory. Its not very sophisticated tech.

>>105619826 (OP)
>Talks relentlessly about his AI company with NO DEMO
>John Carmack fell off
People saw this coming when he made that tweet defending that AI slop test from Microsoft of an “AI generating a level” when level design is directly rooted in architecture principles with psychological models of what the game is supposed to be. The slop Microsoft generated was a random fucking mess and this guy defended it, that obviously means he wants to get rich off AI too. Remember his rocket ship business when Elon started to get popular?

>>105622514
>Remember his rocket ship business when Elon started to get popular?
No lmao, is that real? That should forever discredit the fag

>>105620792
i literally did that wojak face, yikes

>>105620142
There are no geniuses. Just people that spend time learning difficult things while you made excuses.

>Nerds develop a cult of personality around another nerd
>That nerd is actually, surprise surprise, just a nerd in the right place at the right time
Same thing with linux desu.

>>105622775
Sorry, Nutella mate. Not updooting to Windows 11.

>>105622611
There absolutely are geniuses. But yes, there's no such thing as a genius who immediately understands everything always. This is where many genuinely talented people give up. Genius is probably most accurately described as being the confluence of extreme natural intelligence and the even more extreme grit required to stick through hard things after a lifetime of understanding less-hard things immediately.

>>105620376
>deepmind atari playing AI but worse
What a boring sludge
Cant he at least put into a different or custom game at least to be novel in some way

>>105622987
From the talk, from the man himself:
"A reality check for people that think full embodied AGI is right around the corner is to ask your dancing humanoid robot to pick up a joystick and learn how to play an obscure video game."

>>105622933
Look up the Polgar experiment, it disproves the whole idea of geniuses. People just want a reason not to try. I've had so many people complain about drawing being too hard, math being too hard, writing being too hard, music being too hard, programming being too hard. Meanwhile I learnt all of these. The hardest for me was drawing, and I still did it after about a year of trying. People just give up right away and then justify it by saying they're not smart enough or whatever. I mean yeah retards absolutely exist but retards are rare. Most people can learn most things.

>>105622547
Except he started his rcoket company 2 years before Elon.
https://www.youtube.com/watch?v=X68Mm_kYRjc

>>105621719
DOOM is a objectively a good game.
I played DOOM I and II this year and I had fun.
The problem of "modern gamer" is that they look at the graphics instead of gameplay. Game can have stunning graphics, but without good gameplay it's not a GAME.

>>105623110
I want to believe you anon, how did you keep yourself motivated to do all that? Personally i feel burnt out doing anything productive more than two hours a day

>>105625687
Everyone does. It's about maximizing those 2 hours I guess.

>>105621113
sometime's a task is just genuinely really hard and someone's failure to solve it doesn't mean they're a retard

>>105619826 (OP)
Poo hands typed this.

>Be poo
>Have poo degree by poo friends
>Dont have anything valuable
>White people fell off

You're literally worthless

>Be poo
>Fail white people because poo
>Completely fail entire planet cause poo
>White people still problem

Your ancestors literally ate shit and molded you in their stomach, before shitting you out.

Input shit, processing shit, output shit.

Do you think Carmack plays video games?

>>105623110
>Look up the Polgar experiment, it disproves the whole idea of geniuses.
look man, If you've ever met a genuinely dumb person or a genuinely smart person, you would never say that, if there are retards, there are geniuses. The amount of time it takes to learn something is what makes a genius, it might only take them a couple weeks to do what took you a year.

>>105625951
He plays vr beatsaber a lot I think, used to post about it a fair bit not that long ago.

>>105622514
>Remember his rocket ship business when Elon started to get popular?
i think a bunch of rich people started rocket companies at the same time. him, elon, richard branson etc.
the only late entry i heard of was jeff bezos

>>105625687
>how did you keep yourself motivated
Depends on the subject. For maths the mental time limit is about 4 hours. Drawing you can keep going all day it's pretty relaxing. Programming for me is about 4 hours as well. I like the pomodoro technique, then you're not focused on success or failure but just doing your 25 minutes and you can count how many pomodoros you did and feel good about it. Ultimately you need to learn how to have fun without worrying about success or failure. If you go on /ic/ you see the best artists are just fucking around while the "serious" people who are all trying to do perfect anatomy and such have the worst drawings. The key to everything I think is to have fun without worrying about success or failure.

>>105627332
Maybe "fun" is the wrong word, I just mean relax. Things don't need to be perfect on the first try.

>>105626280
Disagree. It might take you 2 weeks to do what takes them a year for a different subject. They just had a head start somewhere you're not aware of. Just like you have a head start somewhere else.

>>105619826 (OP)
john carmack is like the eminem of computer programming

>>105627352
this is counter to reality, If you're good at one subject you are likely to be good at everything else, it can even be isolated as a factor known as "G" . It is probably the most well known scientific finding of the last 150 years when it comes to intelligence, and nobody likes it but its true, and it has been replicated countless times.

>>105627703
> General intelligence can be compared to athleticism. A person might be a very skilled runner, but this does not necessarily mean that they will also be an excellent figure skater.

> However, because this person is athletic and fit, they will probably perform much better on other physical tasks than an individual who is less coordinated and more sedentary.

G is not a counter to what I'm saying. That's simply how the brain works. When you acquire knowledge, it's easier for the brain to acquire more knowledge afterward because we form associations with existing information. Like if you know two languages, it's easier to learn a third and a forth and so on. Just like we all have 2 arms and 2 legs our brains all work generally the same way too. Other than some outlier mutations people aren't as different as you think. Learning new things isn't magic, there's good and bad ways to do it.

>>105619826 (OP)
he realized billionaires will just fork over money if you say "ai"
he's fleecing investors.

>>105619826 (OP)
That was literally the most boring thing I every watched
>we trained an ai.. to play an atari 2600 game... in realtime(slow), one game at a time... we used a 1080p 60fps webcam... but it didn't work if the lighting changed... and i couldn't get my laptop to work right... it couldn't read the score.. and wasn't very good.. and it couldn't extrapolate how to play new games from its previous successes...
>may i see it?
>....no

>>105627703
I'll give you a concrete example, look up the world memory championships. They've been going on for 35 years and all the winners have been regular people just like you. Not a single "genius" or self proclaimed "photographic" memory person has ever beaten completely regular people for 35 years. If such superhumans exist, they could have easily done so but it has never happened.

>>105628016
he said they'll iterate on it a bit and release it publicly

>>105621377
Don't even post this salty retard zoomer shit.

>>105628028
i don't even care
the project sounds retarded from the ground up
>dude parallelization is cheating so we're going to completely avoid the technological advantages that ai actually offers to make an incompetely inept ai that will never keep up with anything let alone mimic human adaptation

>>105627998
G correlates with Speed , so I don't see how its not a counter, the higher your G is , the faster you can acquire skills and knowledge, and over time this effect is profound and insurmountable, if a person can learn in 2weeks what you do in a year, what do you think happens when someone like that studies for 10 years?, G is not everything , but does it show that broadly being able to learn faster applies to all areas, not just specific skills.

and it doesn't matter how small the difference is in terms of mutations, 10% of the population has an iq below 80, and 10% has an iq above 120, in terms of capability there is a HUGE difference. thats a significant chunk of the population, not just a few randos. again if you've ever actually had to work with real people in real life this is immediately obvious.

>>105628024
anon a bunch of obsessed autist spending everyday for years practicing to remember thousands of digits of pi, are not an example of anything other than a tremendous effort. and waste of life. Im pretty sure most people in general have better things to do with their time,

>>105628070
I'm talking about how we're wired biologically. Everyone digests food approximately the same way. Everyone's brains functions about the same way. There are too many environmental factors to take into account to just slap a label on something like IQ or G or whatever else and say that's it. The education system is shit. Most people don't even use their brain the way it's meant to be used so of course you'll not get the results you want to get.

>>105628124
Hardly years, the techniques are built on how the brain is already structured. Understanding the method of loci takes minutes, gathering some locations might take you a day, a week of practise and you can get 100% in all your exams with very little effort. Cause everyone else who isn't aware of how the brain actually works is basically peddling a tricycle at school while people that understand how the mind and brain function are driving a sports car. It's not even close. And it all comes down to just knowing this exists.

>>105628124
Basically you're arguing there's some hardware reason for some people being dumb while I'm telling you it's 90% of the time a software problem. Update the way you see the world and your intelligence will change dramatically.

>>105621315
older guys are getting on testosterone now, and instead of going to their 25 year old self, they choose to go beyond that.

>>105620030
N

>>105628154
Its just not true, nothing supports that assertion,
1. if a person has "good" nutrition and upbringing as in any average western nation, genetics account for 80-90% of adult intelligence,
2. total amount of education/instruction doesn't correlate with intelligence nearly as much as you think it does

the us military tried this,(macnamara's morons), and what they found that people below an iq of 80 cannot be trained to do anything useful in the military, not exactly a high bar, and thus it is illegal to induct such people. these are not people who aren't able bodied, they just aren't intelligent

>>105628165
You can try to cram as much stuff into a person's brain as you can, make your learning as efficient as possible ,but the fundamental limit for how much you can learn is how fast you learn, because humans have a limited lifespan, It limits the level of expertise you can achieve in a human lifespan. This is about as clear cut as it gets.

>>105628293
>what they found that people below an iq of 80
So what that's like 10% of the population. Like I said, 90% of the problem is software not hardware. So unless you were dropped on your head as a child, you can learn most things.

>>105628293
>the us military tried this,(macnamara's morons),
So I looked it up and apparently most of those morons came from low income, unstable families with bad education. The low IQ ones were like 70 IQ or less, basically mentally disabled retards. Which is such a tiny portion of the population I don't know why you're using that as an example. There were way more problems with those troops than only their IQ.

>>105621600
>peaked in the 90s
That's just semantics. Many tech bros also made in 90s, many in early 00s. I wasn't focusing on him specifically but the notion in general.
Autist kun

>>105625583
The problen is that you're a zoomer so you've never played good games in your life

>>105619826 (OP)
Based talk, was wondering what Sutton and Carmack were up to. Keen and AMII have the right philosophy to tackling machine intelligence.

>>105623110
You will never be world class at any.

I never listen to people with assburgers because they don’t care about anyone else but themselves

>>105628334
>>105628334

10% of the population has an iq of <80, the threshold for mental retardation is 70, and let me tell you someone with an iq of 85-90 isnt gonna be learning much, so really your looking at closer to 20% of the population that is virtually untrainable for anything beyond a janitor. and take 90-110 and your maximum aspiration still in the realm of a carpenter to maybe a guy who can do accounting , you cannot train yourself to be a genius , and there is nothing wrong with that, because frankly spending 4x as much time trying to learn something that is beyond your capabilities is a waste of your life.

i absolutely hate what has been done to the education system under this premise, telling people to train for jobs they will struggle with , taking out huge loans and wasting an extra decade of their lives. The people who run this scam should be hanged.

>>105629742
I'm already world class at being able to mix multiple things into successful products. Specialisation is for ants.

>>105629795
>you cannot train yourself to be a genius
Again. Polgar experiment proves you literally can (for regular, non-retarded people).
IQ itself is a questionable metric. You can turn anything into a Gaussian distribution, it's done that way for convenience not because it reflects reality. How many drooling retards do you actually see when you go outside? I mean people that can't tie their shoelaces. How many people at your school couldn't do 2 + 2? I bet it wasn't many, if any.

>>105629231
>I am somehow a zoomer for playing and enjoying a "boomer shooter"
bad reading comprehension or major brain retardation

>>105620540
ya it's pretty unreal. has really exposed midwits too. people every day talking to AI for programming and health advice and ending their chat with "thanks!"

>>105625583
>The problem of "modern gamer" is that they look at the graphics instead of gameplay.
Game devs have been making shitty but pretty-looking games since the NES.

>>105620687
people dont realize how little time it takes to lift weight to the point that you will gain muscle from it. literally in the time it took to type this comment I could have done 1 set with ~25 reps to the point of taxing my muscle and stimulating some minor growth.

>>105619826 (OP)
He fell off when he fired for John Romero for being too much of a gamer to care for the grift

>>105621315
I think he just stopped eating pizza all the time.

>>105622775
I don't think anyone believes Torvalds is a genius. I doubt he considers himself a genius.

>>105620142
>Either you're infallible authority on every subject forever or you're a fraud.
4chan shares a lot in common with Steve Jobs by the looks of it.

>>105620540
I can understand Carmack being excited at new toys, but I just don't see where Tao is pulling the "better than graduate students" shit from. Maybe the big boys have access to better math models but I'm doubtful.