How do you feel about the fact nvidia is abandoning games in favour of AI?
Do you think AI will end up being good for vidya?
How do you feel about the fact nvidia is abandoning games in favour of AI?
Oliver Price
Carson Lewis
Give me an AI gf
Oliver Morgan
You can't fuck a computer
Ethan Russell
Don't tell me what I can't do
Gavin Sanchez
>How do you feel about the fact nvidia is abandoning games in favour of AI?
They are not abandoning anything, that's just so happened that GPU architecture is perfect for deep learning, it would've been stupid not to explore that area for them.
>Do you think AI will end up being good for vidya?
Kinda, sorta, theoretically - machine learning is still quite an underdeveloped area. You can tune algorithm for insane graphic optimisation, genius AI opponent who will rekt any human player, or sort of self-making game, like No Man's Sky, but, you know, working one. However now it's an overkill for gamedev - why the fuck would you need complexity in an industry where everything revolves around design and user experience.
Gabriel Perry
>abandoning games
Source: your ass
Robert Perez
I don't care about shitty companies being desperate.
Intel claws are closing on nvidias neck and i'm having fun watching.
They even tried to shill amd's cpu with their best gpu as "the ultimate desktop pc" in teir desperation
Ayden Robinson
Source: Everyone. PC games don't make money, AI does. Look at the VR """""""revolution""""""""
Levi Davis
>PC games don't make money,
Nvidia Revenue by Markets
Q2 FY17 (miliony $)
Gaming -781
Professional Visualization - 214
Datacenter - 151
Auto - 119
OEM & IP - 163
Total - 1,428
gaming % - 54%
Sure, no money at all, only 54% of total revenue is gaming.
Nolan Lewis
Not with that attitude you pillock.
Cooper Fisher
anandtech.com
>Consequently, the specifications for Drive PX Pegasus are equally forward looking: the board features two unannounced post-Volta next-generation discrete GPUs
THANK YOU BASED NVIDIA
Justin Morgan
>car with HDMI out so you can connect monitor
>suddenly strongest GPU on the market
Elijah White
How do you propose to fuck a computer
Aaron Morris
I hoped we wouldn't achieved AI until we fully undertood how the human brain works, and we had a pretty nice level of nanotechnology advacement (2070 or so), so I'd be death before the apocalypse.
Now I realised I will live to see how a fucking computer kill us all, and I won't be able to do shit about it.
Leo Kelly
Why would a computer kill us?
Grayson Torres
crap its just like IBM and laptops or corning and cookware
>be the best at thing
>no one can beat you at thing
>end up ascending to another tier of thing
>leave everyone with shit products because your competitor who was shit now has no reason to even try
Xavier Bennett
>They are not abandoning anything, that's just so happened that GPU architecture is perfect for deep learning, it would've been stupid not to explore that area for them.
Of course they are not abandoning everyhting, but it doesn't take a business genuis to undertand that they are on the GPU market right fucking now mostly because of AI than games. The company grew 6 times in the last year because of the AI. It's one of the fastest growing IT companies in the worlds of all time because of AI. Their AI-specific tech is leaps ahead of their competitors. It will be a matter of time before they become an AI research facility who happens to produce gaming GPU in addition to their AI chips than a gaming oriented GPU manufacturer
Daniel Martin
What the fuck can AI ACTUALLY do though?
Grayson Davis
You are the fucking reason we don't have AI already. Fucking normies thinking "but muh terminator movies will happen if the computers can think on their own" no it fucking wont because what fucking reason would a computer have to kill us? We have provided computers with life, and we continue to provide computers with everything they need to continue to function, such as new parts and power.
Them going terminator on us would be like a seriously injured patient shooting up a hospital. It's stupid, and achieves literally nothing but self-destruction. Something no self-aware AI would willingly do without an overwhelming reason
Matthew Fisher
>wouldn't achieved AI until we fully undertood how the human brain works
We can't really. How do you make artificial intelligence if you don't even fully understand what the genuine thing is? At best it would be a crude imitation of the parts of the brain we understand thusfar
Lincoln Bennett
I WANT AN AI GF
Connor Scott
A force so stupid that it put retinas backwards and makes nerves to throat take a detour around the heart (recurrent laryngeal nerve) resulted in human intelligence. Moreover, while contemporary Chess engines for example use human-created rules (and as of late, that has been the main driving force behind their improvement, not hardware: modern engines would presumably beat Deep Blue with pawn handicap, while running on more modest hardware), self-taught neural networks can already do specialist tasks at superhuman level without anyone being able to decipher why exactly connections ended up having the weights they did. Consequently, general intelligence also should be well within reach of self-taught or supervised learning systems.
Besides, we DO know a lot of things about ways of thinking, in a lot of aspects well beyond human means. Statistics, decision theory etc. provide us with tools to evaluate truth values of statements or make mathematically best decisions, but just like humans can't use actually efficient sorting algorithms due to minuscule working memory, crunching the numbers is too slow so we have to rely on heuristics known to be flawed. Now, even a superintelligence would have to use heuristics to a degree (not necessarily the ones brain uses, but the best known ones), but even if best modes of reasoning had already been discovered or implemented in the human brain, an artificial intelligence can have a critical advantage in making use of the best methods "intuitively", on top of all of the other advantages being a machine grants by default (such as being immortal, and being able to continue learning while human competitors eventually start losing their wits and ultimately kick the bucket).
Jose King
>self-taught neural networks can already do specialist tasks at superhuman level without anyone being able to decipher why exactly connections ended up having the weights they did
This is the scary shit 2bh. It's all very well to create a program where you understand exactly what is going on with it, but when you can't actually explain why something is happening it goes into a whole other realm.
Robert Bell
GPUs are their cash cow and lack a high % growth because they're already a leading producer in a large market. They won't dial it back at all.