It's more baffling to me that such a smart person wasted so much time on VR.
I'm sure he accomplished interesting technical innovations at Meta but VR as a viable product is a literal meme.
>Is he right
God yes.
except for the last 15 years
He's mostly right. The reason is that you need to develop the expertise to be an expert. This requires years which he just doesn't have. In addition, when you become older, the brain has an increasingly harder time altering memory. This is purported to be a learning-doing tradeoff (you remember clearer, more static things but can't memorize new things as easily).
He's vaxxed to shit lmao. He's dead (pun intended) wrong.
I know you're probably trolling but him being vaxxed is extremely disappointing. It was an IQ test and anyone who took those things when companies like Pfizer have been fined billions over the years is just pure stupidity. It only reinforces the idea that these kinds of people may be brilliant programmers but lack common sense and knowledge outside of their narrow domains.
>It was an IQ test
Wrong. It was a compliance and fear test. Plenty of high IQ idiots took the vax, as did midwits and low IQ retards. The only thing that non-vaxx takers have in common is distrust in government/big companies. Possession of a distrustful nature has more to do with the qualities of one's character than it does the capacity to know various details at once. My brother is a complete fucking retard, but did not take the vaccine.
so he is married to a 4/10 CCP mandated girlfriend, and he is stealing all the time suspicious looking USB sticks from his employers? i wouldn't hire that guy
idk why some people here talk as if he achieved nothing at Meta before quitting, the Quest 2 is a genuinely good very good product (probably the best headset made so far) and so was the Go
Has he done anything noteworthy since stencil shadows in Doom 3? Look like he gradually faded in significance, while he's archrival and nemesis Uncle Tim ate his breakfast and became a billionaire.
Actually VR is developing at steady rate, soon(tm) COTS VR glasses would start reaching human vsion acuity limits and and it will promises intersing thing for remote vision applications.
>VR is even more of a dead end than 3D computer monitors were.
Can you name a single company that still is working on new 3D monitors?
Because I can name plenty that is still working on VR.
"advances in VR" are largely irrelevant to the world when VR remains market nonviable
oh good job he gave the military slightly better tech which will be stolen by china in 1 year, so fucking what? who cares?
>Has he done anything noteworthy since stencil shadows in Doom 3?
Megatextures in the rage engine were a pretty cool idea, I think several other games ended up copying it
Also yes literally in my post that you responded to, the fucking Quest 2 is great, even people who don't like VR that much respect it
Has he done anything noteworthy since stencil shadows in Doom 3? Look like he gradually faded in significance, while he's archrival and nemesis Uncle Tim ate his breakfast and became a billionaire.
does he have to innovate every single year until he's dead? he's not 20 anymore, the mind dulls with time, but the dude is a real genius in his field and has pushed the industry forward several times
It's more baffling to me that such a smart person wasted so much time on VR.
I'm sure he accomplished interesting technical innovations at Meta but VR as a viable product is a literal meme.
Mind doesn't necessarily dull especially at Carmak's age but ambition, energy, and willingness to spend your life in front of a computer screen usually decline somewhat.
Carmack was largely responsible for the asynchronous timewarp stuff in VR. Correct me if I'm wrong, but I believe it's a rendering technique that modifies a frame that's already been drawn instead of drawing an entirely new frame. Saves on overall rendering time because the hardware is drawing fewer unique frames overall.
That's a very old idea - it was implemented in PixelFlow and later ruined by Microsoft when they tried to package it into the Talisman graphics accelerator in the mid-90s
>the days were a single person revolutionizes an entire industry are over
It is not.
It is over for developed industry.
For new industry such happens, computers are example.
[...]
does he have to innovate every single year until he's dead? he's not 20 anymore, the mind dulls with time, but the dude is a real genius in his field and has pushed the industry forward several times
Good engineers are moved into management or do so themselves by starting their own companies. Kek
Megatextures were a precursor for virtualised geometry. Carmack even talked about it in his 2009 presentation. If he is interested in something, you bet your ass it’s achievable within 10-20 years. Fuck the haters, I have no idea what to exactly they’re complaining about. That 10x super focused engineers actually exist?
He can use public datasets until he has a proof of concept and meta will be more than happy to get his money for their data as always. It's true that they won't share the same kind and quality of data that they have internally, but that's easily reproduced by any jackoff AI startup (talking about things like filtering, data assessment, etc. and not, obviously, deanonymizing and data quantity).
A lot of people care a lot about VR. Just because you aren't able to enjoy it doesen't mean others can't.
>carmack was always richgay from wealthy place
Relevant how? Sounds like a bonus to me. Being able to only care about your field while not having to flip burgers makes you a better programmer.
They are the thought leader, of course they're pushing the envelope, that means pushing newest hardware the hardest you can, and the results show. Later on you can think about making it more affordable to poor people.
im not sure were ai is going but its certainly had a better debut than vr which is nothing and nobody is fucking with, especially mark zuckerbergs meta garbage
For top "brains" of MP it is like 40 years.
But Carmack is not the "brain". He is regular bee worker who does job himself. And such people in MP were very young.
He's mostly right. The reason is that you need to develop the expertise to be an expert. This requires years which he just doesn't have. In addition, when you become older, the brain has an increasingly harder time altering memory. This is purported to be a learning-doing tradeoff (you remember clearer, more static things but can't memorize new things as easily).
he's part of the reason why facebook recorded largest stock loss in history. he was never right about anything. pic related: this is carcuck's genius at work.
no amount of ai memes will get back his credibility.
VR is dumb. Augmented reality is where it's at. Imagine if you had normal sunglasses except they could place virtual objects around your surroundings such as pets or gfs like Joi.
VR and AR will never take off with mainstream consumer acceptance until they don't require wearing anything. Until then they will simply serve specific usages. Such as pilots, military, industrial, etc.
That webm is mostly likely just AR overlay on phone's camera, not "normal sunglasses except they could place virtual objects around your surroundings such as pets or gfs like Joi."
Hate to break it to you but AR sunglasses are a pipe dream because truly novel AR experiences require a powerful processor which means large batteries if you want the glasses to run for more than 20 minutes. We need a revolutionary change in batteries before you get AR glasses. So expect VR goggles that can do AR (with 4K cameras) before we have glasses that do AR.
He's 10 years too late to jump unto the AI craze, and the biggest problem is that he hasn't even started trying to understand it even though he claims otherwise. He's screwed. Not to mention he failed to be of any relevance in his previous non-game ventures (aerospace, VR) so why would he think he could do AI? It would be one thing if he wanted to integrate AI in his games, or if he wanted to make a more efficient AI engine, but he wants to do AI research without even knowing what AI is. There's a reason researchers have PhDs, and while for "doing deep learning" you sure as fuck don't need more than an undergrad degree despite job requirements all demanding a master's or PhD, for research you're going to have trouble without the equivalent training and experience of one (which he sure doesn't have).
Meta was ran by retards why would he stay when they ignored him to make that abortion that is burning billions
You could feed our homeless population with what Zuck pissed away
Really shows that at the end of the day that reptile garden gnome can only steal be it ideas or data and when it's time to create something he fucking sucks
As much as I'd like for him to make true on his promises, after watching the Lex Fridman interview, I'm convinced he knows as much about AI as LTT does about Linux and this is just a way for him to stay relevant without actually having to do anything
But in all fairness, if I had the opportunity to scam a bunch of VCs out of $20m, I'd probably do it, too
What problems do you have with what he said? I don't think he'll magically create one when its been a problem for over half a fucking decade, but his mind seemed in the right place at least.
Oh so you really think it needs to run on 80 GB of VRAM and that it can't be reduced at all. That Stable Diffusion was able to make a model comparable to Dall-E that can run on consumer hardware? Stop pretending like it can't be improved by someone who actually wanted to make it run on local computers fast and not attempt to gatekeep the tech like OpenAI by making sure people can't run it locally. OpenAI has a vested interest in making sure only people like them can run the AI. But nice try OpenAI/Microsoft shill.
8 months ago
Anonymous
Allowing AI to be ran unsupervised (locally) is a recipe for disaster.
8 months ago
Anonymous
Take your meds.
8 months ago
Anonymous
Meds, now.
8 months ago
Anonymous
The weights are highly non-sparse, even if you were to do something retarded like compress them at the cost of insane amounts of compute time to do your rolling compute on the compressed weights (lol), you would still see no gains. Similar idea in terms of redundancy.
Yes you can get reductions by doing weird things like binary neural nets. Yes you can use strategies like model distillation (except not, this only works in very limited cases in practice). No that's not going to be significant enough to make it load on commodity computers, and despite that it will come at a large performance cost.
As a reminder chatgpt is a pure bruteforce product, it does not use a clever architecture or anything. It's purely a product of training dynamics on massive amounts of data, and the total model capacity.
However, people have been looking for better archs since before transformers were a thing. The whole point of transformers is foregoing vram efficiency in exchange for compute, which lets it train and infer much faster, leading to the ability to train on masses of data in reasonable-ish time.
People have worked on compressing transformers for ages as well.
While the promising avenue of research right now are things like sgconv, not transformers (an example of arch search that was started before transformers), transformers popularity has diverted basically all funds in sequence learning toward them, including efficiency updates.
Carmack is also not known for efficiency hacks or impossible optimizations.
8 months ago
Anonymous
John Carmack is famous for his proficiency in game programming and his commitment to efficient code and optimization, including performing optimizations that were once considered impossible. For instance, he optimized the Doom and Quake franchises to run smoothly on limited hardware, pushing the boundaries of what was considered possible at the time. These games were revolutionary in terms of their graphics and performance, showcasing Carmack's exceptional skills in code optimization.
8 months ago
Anonymous
>John Carmack is famous for his proficiency in game programming
Not exactly. He's famous for popularizing (but usually not inventing) numerous little interesting tricks (like carmack's reverse or fast inverse square root or hardware scroll on PC using EGA). >including performing optimizations that were once considered impossible
Lolno. >For instance, he optimized the Doom and Quake franchises to run smoothly on limited hardware, pushing the boundaries of what was considered possible at the time.
HAHAHAHHAHAH MY FUCKING SIDES ARE ON FIRE
Is this what your zoomer influencers teach you, zoomzoom? >These games were revolutionary in terms of their graphics and performance,
LOOOOOOL
Read a book, retard.
8 months ago
Anonymous
>falling for AI bait
8 months ago
Anonymous
It's funny though seeing someone with limited knowledge trying to act more knowledgeable than the world's largest database.
8 months ago
Anonymous
I apologize if my previous answer did not meet the expectations of the person who responded. It is great to see that they have strong opinions and I respect their viewpoint. However, it is worth noting that John Carmack's contributions to the field of game programming are widely recognized and respected in the industry. His work on the Doom and Quake franchises did indeed push the boundaries of what was considered possible at the time, and his optimizations allowed these games to run smoothly on limited hardware, which was a major accomplishment. It is also true that he popularized several interesting tricks and techniques, such as Carmack's Reverse and the fast inverse square root. I appreciate the opportunity to clarify and provide a different perspective.
8 months ago
Anonymous
My apologies if my previous answer missed the mark
To the person who responded, with a comment that was dark
But let's give props to John Carmack's name
His contributions to game programming will forever remain
He pushed the limits with Doom and Quake
Smooth play on limited hardware was his claim to fame
Carmack's Reverse, a trick that's still used
And the fast inverse square root, that's just a tiny sliver of what he proved
In the industry, his work is widely recognized
Respected for what he accomplished, no need for revised
Boundaries shattered, he showed us what's possible
His legacy in gaming will remain unstoppable
He pushed the limits with Doom and Quake
Smooth play on limited hardware was his claim to fame
Carmack's Reverse, a trick that's still used
And the fast inverse square root, that's just a tiny sliver of what he proved
I appreciate the chance to clarify and give a new light
To John Carmack's achievements, that shines so bright
A legend in gaming, his impact will never fade
His name will forever be remembered, an accolade.
8 months ago
Anonymous
its OVER for literally any writers and translators
8 months ago
Anonymous
Forsooth, I beg pardon for mine former reply
That mayhap didst not meet the expectation of the esquire
But harken ye, the name of John Carmack dost resound
In the field of game programming, he is widely renowned
With Doom and Quake, he broke the chains
Smooth play on hardware, his claim to fame doth remain
Carmack's Reverse, a trick still used today
And the fast inverse square root, just a small part of what he doth display
In the industry, his deeds are well-known
And his accomplishments, widely overgrown
Boundaries shattered, he hath shown what's feasible
His legacy in gaming, forever shall be
With Doom and Quake, he broke the chains
Smooth play on hardware, his claim to fame doth remain
Carmack's Reverse, a trick still used today
And the fast inverse square root, just a small part of what he doth display
I give thanks for this chance to set the record straight
For John Carmack's triumphs, that doth shine so great
A legend in gaming, his impact shall not fade
His name shall be remembered, in every minstrel's ballad.
it's really something if youre an imaginative person kek
8 months ago
Anonymous
>his claim to fame doth remain
this is eminem tier
8 months ago
Anonymous
100,000,000 million albums sold?
hell, I'll take it
8 months ago
Anonymous
The Game Hut dude is literally more intelligent than Carmack if we're going by cleverness.
From what I've seen of him Carmack is obviously intelligent but I feel as though his celebrity status outshines his actual technical aptitude. There are so many great programmers that exist that no one pays attention to its insane.
>It was an IQ test
Wrong. It was a compliance and fear test. Plenty of high IQ idiots took the vax, as did midwits and low IQ retards. The only thing that non-vaxx takers have in common is distrust in government/big companies. Possession of a distrustful nature has more to do with the qualities of one's character than it does the capacity to know various details at once. My brother is a complete fucking retard, but did not take the vaccine.
"IQ test" was more of a figure of speech than me literally equating it to an IQ test. It takes critical thinking and focus to be distrustful of governments and corporations when the whole worlds telling you to they're your best friends. Obviously different groups of people have different motivations for taking and not taking them, but generally the people I've seen who didn't take them are more open-minded.
8 months ago
Anonymous
>popularizing
he made a working binary space partitioning implementation with Doom and then Quake
Not him but >nobody else is working on AGI!1!
Oh yeah except some of the largest labs in the world renowned for their numerous breakthroughs in deep learning for several decades
https://github.com/mila-iqia/babyai >I-I read the c-code once and the c-core logic is 1k lines
He's literally mistaking boiler plate for "AI logic", how dumb do you have to be? >OpenAI tried to poach me and I said no and to give me AI papers. I totally read them all you guys
Still shows he doesn't know the first thing about it, with most of his comments suggesting he thinks it's an if-else forest.
That was from another interview, not LF's.
VR requires shit ton of processing power and good inexpensive hardware and neither of that is currently possible there needs a push in AI first before that so he is indeed right
He copied the source code of Rage onto his usb which had code for VR related stuff, and googled how to wipe a usb after he was getting sued by ZeniMax, and obviously lost the case (twice, because he tried to sue ZeniMax back).
If he left he could more easily forget about that.
not anon, but it was his, and employees at occulus snitched on him for basically saying "hey here is the source code for this game, use it as a reference"
AI is clearly going to be the bigger field at this point. ML will be to general purpose computing what general purpose computing was to computers.
VR will continue to have a niche fanbase of gamers but it'll always be another CS subfield. Billionaires will be minted in ML.
VR has really been a retarded pursuit, VR works perfectly since the 70s or 80s, but nobody uses it except 1 time in their lives when they try it at an exposition their parents brought them as kids and never care again
always has been
He's vaxxed to shit lmao. He's dead (pun intended) wrong.
>mass-replies
>doesn't even check what the posts are about
This should be a bannable offense.
BOT should have a login system and karma, also like and subscribe to my posts
No refunds.
Two more weeks.
Till we stop the spread?
Trust the plan.
~~*Q*~~ lmao. Trump was the proud father of the !Vaxx. You can't pigeonhole me with MIGAtards. You will never be fully vaccinated.
I know you're probably trolling but him being vaxxed is extremely disappointing. It was an IQ test and anyone who took those things when companies like Pfizer have been fined billions over the years is just pure stupidity. It only reinforces the idea that these kinds of people may be brilliant programmers but lack common sense and knowledge outside of their narrow domains.
I'm indifferent to Carmack. I just like seeing and causing vax cattle pain. No real malice just euphoric schadenfreude.
Rent free
>It was an IQ test
Wrong. It was a compliance and fear test. Plenty of high IQ idiots took the vax, as did midwits and low IQ retards. The only thing that non-vaxx takers have in common is distrust in government/big companies. Possession of a distrustful nature has more to do with the qualities of one's character than it does the capacity to know various details at once. My brother is a complete fucking retard, but did not take the vaccine.
Yeah. People like to think that distrust implies intelligence as opposed to blind trust, but they're really completely unrelated properties.
>but lack common sense and knowledge outside of their narrow domains.
that much was already apparent
so he is married to a 4/10 CCP mandated girlfriend, and he is stealing all the time suspicious looking USB sticks from his employers? i wouldn't hire that guy
This far Carmack hasn't been wrong
except for the last 15 years
vr is a solid technology, this 1 to 1 mapping to a virtual world hasn't been done before, all previous attempts were around the stereo effect
He's also right about VR, it definitely has a future.
Carmack must be the most retarded genius ever.
it'd been much better if he'd stuck to developing doom 5 or some shit like that.
idk why some people here talk as if he achieved nothing at Meta before quitting, the Quest 2 is a genuinely good very good product (probably the best headset made so far) and so was the Go
Has he done anything noteworthy since stencil shadows in Doom 3? Look like he gradually faded in significance, while he's archrival and nemesis Uncle Tim ate his breakfast and became a billionaire.
Advances in VR are in large part thanks to him
VR is even more of a dead end than 3D computer monitors were.
Actually VR is developing at steady rate, soon(tm) COTS VR glasses would start reaching human vsion acuity limits and and it will promises intersing thing for remote vision applications.
>VR is even more of a dead end than 3D computer monitors were.
Can you name a single company that still is working on new 3D monitors?
Because I can name plenty that is still working on VR.
"advances in VR" are largely irrelevant to the world when VR remains market nonviable
oh good job he gave the military slightly better tech which will be stolen by china in 1 year, so fucking what? who cares?
>Has he done anything noteworthy since stencil shadows in Doom 3?
Megatextures in the rage engine were a pretty cool idea, I think several other games ended up copying it
Also yes literally in my post that you responded to, the fucking Quest 2 is great, even people who don't like VR that much respect it
does he have to innovate every single year until he's dead? he's not 20 anymore, the mind dulls with time, but the dude is a real genius in his field and has pushed the industry forward several times
>does he have to innovate every single year until he's dead?
uhm yes
It's more baffling to me that such a smart person wasted so much time on VR.
I'm sure he accomplished interesting technical innovations at Meta but VR as a viable product is a literal meme.
It's direct step continuing his previous work. 3D graphics, more immersion, more copying of real life by computer.
Mind doesn't necessarily dull especially at Carmak's age but ambition, energy, and willingness to spend your life in front of a computer screen usually decline somewhat.
Carmack was largely responsible for the asynchronous timewarp stuff in VR. Correct me if I'm wrong, but I believe it's a rendering technique that modifies a frame that's already been drawn instead of drawing an entirely new frame. Saves on overall rendering time because the hardware is drawing fewer unique frames overall.
That's a very old idea - it was implemented in PixelFlow and later ruined by Microsoft when they tried to package it into the Talisman graphics accelerator in the mid-90s
https://www.computer.org/publications/tech-news/chasing-pixels/microsofts-talisman
>it's a rendering technique that modifies a frame that's already been drawn instead of drawing an entirely new frame.
the days were a single person revolutionizes an entire industry are over
>the days were a single person revolutionizes an entire industry are over
It is not.
It is over for developed industry.
For new industry such happens, computers are example.
Good engineers are moved into management or do so themselves by starting their own companies. Kek
Carmack literally just quit managing at facebook to start his own research though. He said he wants to have more hands on part in it.
What would it take to get a job at Keen?
>Keen
The guys who made Space Engineers?
No. He named his startup Keen Technologies which is coincidentally the same as a company out of Hyderabad.
Megatextures were a precursor for virtualised geometry. Carmack even talked about it in his 2009 presentation. If he is interested in something, you bet your ass it’s achievable within 10-20 years. Fuck the haters, I have no idea what to exactly they’re complaining about. That 10x super focused engineers actually exist?
>10x engineer
>costs 10000x
Not even remotely true. Not even 10x the cost.
Doesn't he charge $2000 per hour?
He's only worth $50 an hour, but people don't pay for talent or quality, they pay for notoriety; brand name.
Where did you get that from?
google
you don't need 10000 mediocre programs
You need just 1 good program you can copy for everyone
>saves 100000x in the long run.
Turns out software scales
>as if he achieved nothing at Meta before quitting
anything is better than working for facebook.
Ironically, facebook is one of the best companies to work on AI at.
yes, Carmack has to somehow buy the training dataset and Zuck will never share anything. he even took his Beat Saber game away
He can use public datasets until he has a proof of concept and meta will be more than happy to get his money for their data as always. It's true that they won't share the same kind and quality of data that they have internally, but that's easily reproduced by any jackoff AI startup (talking about things like filtering, data assessment, etc. and not, obviously, deanonymizing and data quantity).
>armadillo aerospace
A lot of people care a lot about VR. Just because you aren't able to enjoy it doesen't mean others can't.
>carmack was always richgay from wealthy place
Relevant how? Sounds like a bonus to me. Being able to only care about your field while not having to flip burgers makes you a better programmer.
They are the thought leader, of course they're pushing the envelope, that means pushing newest hardware the hardest you can, and the results show. Later on you can think about making it more affordable to poor people.
im not sure were ai is going but its certainly had a better debut than vr which is nothing and nobody is fucking with, especially mark zuckerbergs meta garbage
>Is he right?
It doesn't matter.
He is too old to do something new good.
That graph is for regular bee workers, not for CTOs
For top "brains" of MP it is like 40 years.
But Carmack is not the "brain". He is regular bee worker who does job himself. And such people in MP were very young.
ok anon
He's mostly right. The reason is that you need to develop the expertise to be an expert. This requires years which he just doesn't have. In addition, when you become older, the brain has an increasingly harder time altering memory. This is purported to be a learning-doing tradeoff (you remember clearer, more static things but can't memorize new things as easily).
Who is this? Is he the Australian guy who tears apart computers?
he's part of the reason why facebook recorded largest stock loss in history. he was never right about anything. pic related: this is carcuck's genius at work.
no amount of ai memes will get back his credibility.
>this is carcuck's genius
no it's not, he said this dumb shit is why he quit
i gotta hear this. what he'd say exactly?
Not who you replied to, but here was the full resignation letter. https://pastebin.com/5knhk06Q
It ends with
>Make better decisions and fill your products with “Give a Damn”!
Carmack is art director of Facebook?
Show me the art director of Facebook.
VR is dumb. Augmented reality is where it's at. Imagine if you had normal sunglasses except they could place virtual objects around your surroundings such as pets or gfs like Joi.
VR and AR will never take off with mainstream consumer acceptance until they don't require wearing anything. Until then they will simply serve specific usages. Such as pilots, military, industrial, etc.
Not even a small band on your neck?
god I cant wait to get an augmented reality AI e-boi GF
>technology we have now is dumb
>technology that we will never have is where it's at
but we do have it, that webm is from like ikea
That webm is mostly likely just AR overlay on phone's camera, not "normal sunglasses except they could place virtual objects around your surroundings such as pets or gfs like Joi."
this webm does not work on my android phone for some reason at all
what gives?
>around your surroundings
why would I want to be in boring reality, when I can be in unimaginable fantasy lands
Remember when Microsoft was working on an AR headset when everyone was starting with VR?
Hate to break it to you but AR sunglasses are a pipe dream because truly novel AR experiences require a powerful processor which means large batteries if you want the glasses to run for more than 20 minutes. We need a revolutionary change in batteries before you get AR glasses. So expect VR goggles that can do AR (with 4K cameras) before we have glasses that do AR.
Carmack is a joke compared to this guy. This guy is literally the Jesus of VR and will bring VR to the masses before he dies.
>Is he right
God yes.
good poc, do something funny now
He's rich and successful enough that he can do whatever he likes and not worry too much about whether it's going to pan out.
He's already made it.
he did an interview were he explains his reasoning, he thinks AGI is close too
https://dallasinnovates.com/exclusive-qa-john-carmacks-different-path-to-artificial-general-intelligence/
>research on VR all together after 15 years
yeah, he's craze
VR wasn't worth even a day of research
He's 10 years too late to jump unto the AI craze, and the biggest problem is that he hasn't even started trying to understand it even though he claims otherwise. He's screwed. Not to mention he failed to be of any relevance in his previous non-game ventures (aerospace, VR) so why would he think he could do AI? It would be one thing if he wanted to integrate AI in his games, or if he wanted to make a more efficient AI engine, but he wants to do AI research without even knowing what AI is. There's a reason researchers have PhDs, and while for "doing deep learning" you sure as fuck don't need more than an undergrad degree despite job requirements all demanding a master's or PhD, for research you're going to have trouble without the equivalent training and experience of one (which he sure doesn't have).
>failed to be of any relevance in his previous non-game ventures
he made the rift CV1 that defined the future of headsets and VR controllers
L-lol?
imagine that, all headsets have to copy the CV1 controller scheme and it still has the lowest lag
(You)
>he doesn't remember the google cardboard and "smartphone vr enclosures"
VR is irrelevant
Meta was ran by retards why would he stay when they ignored him to make that abortion that is burning billions
You could feed our homeless population with what Zuck pissed away
Really shows that at the end of the day that reptile garden gnome can only steal be it ideas or data and when it's time to create something he fucking sucks
https://vocaroo.com/1hMzsDhzxuTG
lacks the uhms in the end of every sentence
>Carmack will invent AGI because... because he just will, ok?
As much as I'd like for him to make true on his promises, after watching the Lex Fridman interview, I'm convinced he knows as much about AI as LTT does about Linux and this is just a way for him to stay relevant without actually having to do anything
But in all fairness, if I had the opportunity to scam a bunch of VCs out of $20m, I'd probably do it, too
>Lex Fridman
What problems do you have with what he said? I don't think he'll magically create one when its been a problem for over half a fucking decade, but his mind seemed in the right place at least.
Carmack might be one that figures out how to make ChatGPT run on a local computer with optimizations. Carmack's one kink is impossible optimizations.
You have no idea whatsoever how this tech works.
Oh so you really think it needs to run on 80 GB of VRAM and that it can't be reduced at all. That Stable Diffusion was able to make a model comparable to Dall-E that can run on consumer hardware? Stop pretending like it can't be improved by someone who actually wanted to make it run on local computers fast and not attempt to gatekeep the tech like OpenAI by making sure people can't run it locally. OpenAI has a vested interest in making sure only people like them can run the AI. But nice try OpenAI/Microsoft shill.
Allowing AI to be ran unsupervised (locally) is a recipe for disaster.
Take your meds.
Meds, now.
The weights are highly non-sparse, even if you were to do something retarded like compress them at the cost of insane amounts of compute time to do your rolling compute on the compressed weights (lol), you would still see no gains. Similar idea in terms of redundancy.
Yes you can get reductions by doing weird things like binary neural nets. Yes you can use strategies like model distillation (except not, this only works in very limited cases in practice). No that's not going to be significant enough to make it load on commodity computers, and despite that it will come at a large performance cost.
As a reminder chatgpt is a pure bruteforce product, it does not use a clever architecture or anything. It's purely a product of training dynamics on massive amounts of data, and the total model capacity.
However, people have been looking for better archs since before transformers were a thing. The whole point of transformers is foregoing vram efficiency in exchange for compute, which lets it train and infer much faster, leading to the ability to train on masses of data in reasonable-ish time.
People have worked on compressing transformers for ages as well.
While the promising avenue of research right now are things like sgconv, not transformers (an example of arch search that was started before transformers), transformers popularity has diverted basically all funds in sequence learning toward them, including efficiency updates.
Carmack is also not known for efficiency hacks or impossible optimizations.
John Carmack is famous for his proficiency in game programming and his commitment to efficient code and optimization, including performing optimizations that were once considered impossible. For instance, he optimized the Doom and Quake franchises to run smoothly on limited hardware, pushing the boundaries of what was considered possible at the time. These games were revolutionary in terms of their graphics and performance, showcasing Carmack's exceptional skills in code optimization.
>John Carmack is famous for his proficiency in game programming
Not exactly. He's famous for popularizing (but usually not inventing) numerous little interesting tricks (like carmack's reverse or fast inverse square root or hardware scroll on PC using EGA).
>including performing optimizations that were once considered impossible
Lolno.
>For instance, he optimized the Doom and Quake franchises to run smoothly on limited hardware, pushing the boundaries of what was considered possible at the time.
HAHAHAHHAHAH MY FUCKING SIDES ARE ON FIRE
Is this what your zoomer influencers teach you, zoomzoom?
>These games were revolutionary in terms of their graphics and performance,
LOOOOOOL
Read a book, retard.
>falling for AI bait
It's funny though seeing someone with limited knowledge trying to act more knowledgeable than the world's largest database.
I apologize if my previous answer did not meet the expectations of the person who responded. It is great to see that they have strong opinions and I respect their viewpoint. However, it is worth noting that John Carmack's contributions to the field of game programming are widely recognized and respected in the industry. His work on the Doom and Quake franchises did indeed push the boundaries of what was considered possible at the time, and his optimizations allowed these games to run smoothly on limited hardware, which was a major accomplishment. It is also true that he popularized several interesting tricks and techniques, such as Carmack's Reverse and the fast inverse square root. I appreciate the opportunity to clarify and provide a different perspective.
My apologies if my previous answer missed the mark
To the person who responded, with a comment that was dark
But let's give props to John Carmack's name
His contributions to game programming will forever remain
He pushed the limits with Doom and Quake
Smooth play on limited hardware was his claim to fame
Carmack's Reverse, a trick that's still used
And the fast inverse square root, that's just a tiny sliver of what he proved
In the industry, his work is widely recognized
Respected for what he accomplished, no need for revised
Boundaries shattered, he showed us what's possible
His legacy in gaming will remain unstoppable
He pushed the limits with Doom and Quake
Smooth play on limited hardware was his claim to fame
Carmack's Reverse, a trick that's still used
And the fast inverse square root, that's just a tiny sliver of what he proved
I appreciate the chance to clarify and give a new light
To John Carmack's achievements, that shines so bright
A legend in gaming, his impact will never fade
His name will forever be remembered, an accolade.
its OVER for literally any writers and translators
Forsooth, I beg pardon for mine former reply
That mayhap didst not meet the expectation of the esquire
But harken ye, the name of John Carmack dost resound
In the field of game programming, he is widely renowned
With Doom and Quake, he broke the chains
Smooth play on hardware, his claim to fame doth remain
Carmack's Reverse, a trick still used today
And the fast inverse square root, just a small part of what he doth display
In the industry, his deeds are well-known
And his accomplishments, widely overgrown
Boundaries shattered, he hath shown what's feasible
His legacy in gaming, forever shall be
With Doom and Quake, he broke the chains
Smooth play on hardware, his claim to fame doth remain
Carmack's Reverse, a trick still used today
And the fast inverse square root, just a small part of what he doth display
I give thanks for this chance to set the record straight
For John Carmack's triumphs, that doth shine so great
A legend in gaming, his impact shall not fade
His name shall be remembered, in every minstrel's ballad.
it's really something if youre an imaginative person kek
>his claim to fame doth remain
this is eminem tier
100,000,000 million albums sold?
hell, I'll take it
The Game Hut dude is literally more intelligent than Carmack if we're going by cleverness.
From what I've seen of him Carmack is obviously intelligent but I feel as though his celebrity status outshines his actual technical aptitude. There are so many great programmers that exist that no one pays attention to its insane.
"IQ test" was more of a figure of speech than me literally equating it to an IQ test. It takes critical thinking and focus to be distrustful of governments and corporations when the whole worlds telling you to they're your best friends. Obviously different groups of people have different motivations for taking and not taking them, but generally the people I've seen who didn't take them are more open-minded.
>popularizing
he made a working binary space partitioning implementation with Doom and then Quake
Are you retarded?
you think Doom had BSP from the beginning?
Damn, you really are retarded.
Not him but
>nobody else is working on AGI!1!
Oh yeah except some of the largest labs in the world renowned for their numerous breakthroughs in deep learning for several decades
https://github.com/mila-iqia/babyai
>I-I read the c-code once and the c-core logic is 1k lines
He's literally mistaking boiler plate for "AI logic", how dumb do you have to be?
>OpenAI tried to poach me and I said no and to give me AI papers. I totally read them all you guys
Still shows he doesn't know the first thing about it, with most of his comments suggesting he thinks it's an if-else forest.
That was from another interview, not LF's.
>$20m
>a lot
that's pocket change his billionaire friends threw at him because they feel bad for him
Carmack goes where interesting new things to learn and then improve are. It's why he eventually left gaming for aerospace and VR.
Its just opportunism. In the LF interview he said that he would've worked on nuclear energy if he hadn't gone down the AGI path.
Kind of sounds like he went where there are interesting new things in an attempt to improve them.
yes but idk what he can do there
VR requires shit ton of processing power and good inexpensive hardware and neither of that is currently possible there needs a push in AI first before that so he is indeed right
He copied the source code of Rage onto his usb which had code for VR related stuff, and googled how to wipe a usb after he was getting sued by ZeniMax, and obviously lost the case (twice, because he tried to sue ZeniMax back).
If he left he could more easily forget about that.
>and googled how to wipe a usb
just throw it in the toilet? is he retarded?
maybe this usb was corporate property
why did he take it? couldn't afford one?
not anon, but it was his, and employees at occulus snitched on him for basically saying "hey here is the source code for this game, use it as a reference"
>is he retarded?
mmm
Fine, it's your funeral
AI is clearly going to be the bigger field at this point. ML will be to general purpose computing what general purpose computing was to computers.
VR will continue to have a niche fanbase of gamers but it'll always be another CS subfield. Billionaires will be minted in ML.
not computers, calculators
tbh he has been irrelevant since leaving id
he's smart but he's just wasting his talent for gimmick shit
VR has really been a retarded pursuit, VR works perfectly since the 70s or 80s, but nobody uses it except 1 time in their lives when they try it at an exposition their parents brought them as kids and never care again
if you're going to post continuously in this thread for days could you make your english a little better
it's extremely obvious you are ESL and it makes all your posts very recognizable
i posted 3 times in this thread and only 1 related to VR, can you guess the other 2? the game commences now