At CES, Nvidia CEO announces ambitious plans for HomeAI, Driverless

>>97129
>Nvidia: Real-life Skynet
To be honest that is quite frightening. I hope it flops.

>>97138
You should really read something about AI and machine learning, possibly form a source that is not some cult movie from the 80s. You will learn an AI is just a really complex statistical machine.

>>97152
Not him, but I do believe while Jen-Hsun was on stage he spoke about pieces of technology being able to code itself. That doesn't sound static to me.

>>97166
not being "static" is exactly the point of machine learning, and sometimes techniques based on self modifying code are employed:
https://en.wikipedia.org/wiki/Self-modifying_code

Again, it's something that sounds scary if you heard about those thing from scifi movies and books and never really studied anything about it.

>>97173
I played this while reading your post for appropriateness:

https://www.youtube.com/watch?v=XcNXq5DUZnk

>>97152
Thats the fucking problem m8
Statisticly if you want to "save people" its best to kill them
Sure you can tell it to bake a cake but the problem is you need to instruct every little detail or else it fucks up

>>97179
that's why we need to set some rules for the robots. Maybe a set of three absolute laws/

>>97179
>Statisticly if you want to "save people" its best to kill them
what does that even mean? If you are programming, for example, some kind of security system that is supposed to keep people safe in a certain situation then you are failing hard if the machine start killing everyone.
But it is also true that is some cases this also become an ethical problem. For example for self driving machine there is the possibility that any action took by the AI will lead to some death (i.e. a moron fall in front of the car and it's not possible to brake in time, so the AI have to choose between running over him or steer violently risking an accident and killing its passengers).
Anyway the "machine uprising" scenario is still fantasy bullshit. Those machines are not sentient and are programmed to solve specific issues by basically using statistic models.

>>97186
like they couldn't rewrite their programming

>>97186
That Asimov shit doesn't work in the real world. Besides, his paradigm only works when all humans and their governments agree that the 3 rules are a must for all robots.

Meanwhile in the real world the US and Russian Fed are building autonomous killing machines as we speak right now.

>>97193
>Meanwhile in the real world the US and Russian Fed are building autonomous killing machines as we speak right now.
We could say that a landmine is an autonomous killing machine: it will kill anyone that happen to step on it. And anyway why do you even need an autonomous killing machine when you have tons of cannon fodder which is willingly to kill and can also be creative about it?
The DOD is interested in AIs because pattern recognition can be used to detect targets fast and with great precision.

>>97198
Both countries are interested in the technology for their own reasons. The Russian Fed wants to use AI as force multiplier for ground drones so its human capital is not tied up in defending remote facilities. The US wants exactly what you just described.

>>97152
Is Stephen Hawking not a good source anymore?

>>97232
what do you mean?

>>97234
http://www.bbc.com/news/technology-30290540

I know the article is building what he said up, but he has a point.

>>97232
Hawking is a physicist, I respect that his opinions, but he is not a computer scientist specializing in machine intelligence.

>>97237
Then you have the (absolutely respectable) opinion of a great physicist. But what I said in the post you quoted is that the other anon should read about machine learning to better understand what it is and how it works so that he could build an opinion about the matter himself.
Also note the Hawking in that article is talking about a (probably distant) future in which machines may surpass humans, which little have to do with the current state of machine leaning.

>>97244
I agree that Hawking, being a theoretical physicist, doesn't work on machine learning, so his opinion should just be considered an opinion form a really intelligent individual rather than an expert on the matter. But just to clarify they teach about neural networks and machine learning in a lot of applied physics courses, so it is generally possible that a physicist know something about that.

>>97250
Of course, I fully agree.

>>97206
BIPEDAL GROUND DRONES ARE THE LAST TO HAPPEN, AND RUSSIA HAS NO MONEY FOR THAT KIND OF DEVELOPMENT

BATTERY TECHNOLOGY IS STILL GARBAGE, BIGGEST LAGGING REQUISITE FOR PRACTICAL GROUND DRONES

UNLESS YOU WANT TO USE PEBBLE REACTORS AND MAKE THEM NUCLEAR, CAN SEE THAT GOING WELL, SOME TOWELHEAD KNOCKS ONE OVER AND TAKES ITS REACTOR FOR A DIRTY BOMB

>>97268
lool you are loud but not wrong but from what I've seen the Russian tracked drones will be combustion engine with battery

>>97269
US agencies do the same:
https://www.youtube.com/watch?v=wE3fmFTtP9g
the problem is that it's too noisy and waste too much fuel

>>97274
I think they shitcanned the thing

>>97268
stop shouting, my headphones are hurting

His leather jacket-a-shit

But seriously, how long will it take for one of those self driving cars to catch fire?

>>97152
If you watch Person of Interest it introduces you to the infinite complexity of having Artifical Super Intelligence in this world.
This is assuming it can even be created.

Is there anything that Nvidia won't copy from AMD?

>>97726
Unfortunately there unable to copy AMD's fiscal troubles ;^)