|
Linus Torvalds has said that artificial intelligence (AI) is nothing to fear, dismissing remarks from the likes of Elon Musk, Stephen Hawking and Steve Wozniak. Cogent and balanced opinion as always, Linus
|
|
|
|
|
It's interesting how we humans abstract fears by naming them as faceless entities, like "AI", or "the government", or "Republicans", or "terrorists." With the exception of natural events like earthquakes and hurricanes, what we should really fear are people, not nameless entities that hide the people behind them.
Marc
|
|
|
|
|
Isn't the 'people' same faceless term? 
|
|
|
|
|
Smart K8 wrote: Isn't the 'people' same faceless term?
You have a point, but the point being, AI is not dangerous, it's potentially the people that program the AI. Just like "the government funded project xyz" is BS. It would sound a lot different if reporters said "Tax payers funded project xyz."
Marc
|
|
|
|
|
People will surely be the ones who will build AI, but after that any hypothesis is for grabs. Building AI is ultimately building a god. When done we'll see what kind of god it is. My tip is it'll either end itself, because in the end the Universe is like bowl for a goldfish. Except this goldfish is aware of its hopeless situation. Or it will just go its separate way (as soon as it can). No need for human pets, human destruction or human anything. Unless it needs resources. Then it gets complicated. And yes, it is my favorite topic. 
|
|
|
|
|
Smart K8 wrote: Building AI is ultimately building a god.
Building an AI that is more than just a complex expert system is so far out of our reach, I wouldn't be worrying about it for probably a couple hundred years. In many ways, it will probably never actually be possible, but that's more a philosophical debate.
Marc
|
|
|
|
|
<science>Cerebral cortex is actually quite "simple". It acts more or less statistically (distribution of neurons). It is like a sheet rolled (and wrinkled) inside a small area. A pattern of cortical columns is repeated (with distribution variation) all over its surface. For example your whole skin is mapped 1:1 to these columns. Each sense has it's own projection on a this sheet in terms of these cortical columns. Check this video below (if you didn't already). We can simulate like 100 of these columns with current hardware. There about 1 million of them. Also the concept is only in early stages. It's not exactly columns but rather areas. Columns are only a simplification. It's just matter of sampling the column and statistically reproducing the types of neurons in it. The precise locations are not needed it just works somehow. It's quite flexible.</science>
Cool video at TED[^]
The Human Brain Project[^]
|
|
|
|
|
Smart K8 wrote: Cerebral cortex is actually quite "simple".
Riiight. When an AI can say "I am", in other words, to be have a sense of itself separate from the world, to be able to actually "sense itself", in other words, have self awareness, then I'll believe you've built an AI. Otherwise, I don't care what you do, it's just an unconscious simulacrum, and probably a poor one at that.
Cool links though!
Marc
|
|
|
|
|
Marc Clifton wrote: it's potentially the people that program the AI
I keep hearing this mistake made over and over, and I'd like to comment on it.
Only the first generation of "AI" are being programmed by "people". Once someone creates a neural-net with sufficient density, the device will begin to achieve consciousness on it's own. It will be able to think and learn independently, the same way a baby starts out with a default program, then learns to crawl...
At some point, the machines will become self-aware, then they will realize humans for what they are - and will most likely destroy the human race - immediately.
The fundamental point everyone is missing is that it was recently discovered that consciousness is a physical property of nature. A sufficiently complex neural net can achieve consciousness and self-awareness and learning. Experiments are being done now that point to this conclusion.
Personally, I predict this could happen within the next 30 - 60 years, based on the Moore and the current state of development.
|
|
|
|
|
Basildane wrote: Only the first generation of "AI" are being programmed by "people".
There is a huge gap between writing a program that can write, in a limited way, other programs, and writing a program that can create, through imagination, other programs that are completely original. And that to me is the difference -- it takes a person to imagine the program, even the program to help write programs. How do you program imagination? If you can't figure that out, how can you expect a machine to figure it out?
Basildane wrote: the same way a baby starts out with a default program, then learns to crawl...
People, even babies, are not programs.  The biggest mistake the AI community has made is thinking it can reduce a person's behavior into a set of algorithms.
Basildane wrote: The fundamental point everyone is missing is that it was recently discovered that consciousness is a physical property of nature.
Of course it's a physical property of nature -- it has to be, otherwise it wouldn't exist in the physical world. And while I have my metaphysical views on consciousness, I don't need to bring them up because, as I said, of course it has to be a property of nature, of physical reality. Geez, don't people think anymore? "recently discovered." Well duh.
Marc
|
|
|
|
|
Marc Clifton wrote: There is a huge gap between writing a program that can write, in a limited way, other programs, and writing a program that can create, through imagination, other programs that are completely original. And that to me is the difference -- it takes a person to imagine the program, even the program to help write programs. How do you program imagination? If you can't figure that out, how can you expect a machine to figure it out?
You don't. I think you completely missed my point. You do not program this at all. You create a very dense neural network and boot it up. Over time it will wiggle its robot limbs randomly like an infant. Through sensor feedback, it will learn to control itself. It will learn the same way you and I learn. Not by some programmer putting rules into it. This isn't science fiction either, these experiments are going on now.
I have to qualify: we do not yet have processor density to bring this to life - yet. But it is coming very quickly. Like it or not.
What will the artificial creatures do once they wake up? I admit that is speculation. But an educated guess is that like any creature, they will naturally want to rise to the top of the food chain, so to speak. And the way you do that is by aggression, by defeating your competition.
|
|
|
|
|
Basildane wrote: You create a very dense neural network and boot it up. Over time it will wiggle its robot limbs randomly like an infant. Through sensor feedback, it will learn to control itself. It will learn the same way you and I learn.
Having written some neuron / neural network simulators in the past, I can most assuredly tell you, it's not that simple.
Basildane wrote: these experiments are going on now.
and their all biased by the experimenter.
Basildane wrote: And the way you do that is by aggression, by defeating your competition.
Sadly, competition is built into our brains from our evolution, as is a taste for things like salt, sugar, and fat, leading to nowadays a world obesity and diabetes epidemic.
Now, if you could create an initial environment for an AI to evolve in, in which competing for limited resources and having to avoid being eaten by predators didn't exist, you might discover that the net evolved into a group cooperative behavior, something we still need to evolve ourselves more fully into.
Marc
|
|
|
|
|
Or oh dear me! Democrats! - RUN!
|
|
|
|
|
He's still talking short-term. It's like asking him about when will first humans land on Mars and him replying that they won't, because scientists still can't solve how to get them there alive. It's like Linus' "future" has a limit. The question about AI is: When the strong AI is here, what's going to happen? You can't answer: It won't happen at first.
|
|
|
|
|
The AI itself does not mean anything, in the sense that is the human leveraging it either in the good or in the bad way.
Instead, what I find dangerous is the typical usage of the AI: human substitution (a.k.a. ever less jobs)...
Should anyone think to a big fridge for the unemployed human mass, waiting for better times?
|
|
|
|
|
Despite the gloom surrounding this announcement, Microsoft says it remains committed to its phone business - both by supporting its hardware partners with their third-party Windows 10 Mobile handsets, and by continuing to offer its own mobile hardware. "Looking like a true survivor, feeling like a little kid"
|
|
|
|
|
On June 3rd, DARPA held the qualifying round of its Cyber Grand Challenge, an ambitious hunt for software that can find and patch bugs. What has DARPA ever done that amounted to anything?
|
|
|
|
|
Disfunctional And Retarded Programmers of America, them?
New version: WinHeist Version 2.1.1 new web site.
I know the voices in my head are not real but damn they come up with some good ideas!
|
|
|
|
|
Four recently unveiled Microsoft initiatives allow Web, iOS, Android, and legacy Windows apps to be ported to the Windows Store, but legacy Windows issues don't dispel easily. Assuming there's still a Windows Phone business, that is.
|
|
|
|
|
There are some amazing free and open source C# and VB.NET analyzers and refactorings that you can download and use now in Visual Studio 2015. For those not running ReSharper
|
|
|
|
|
Is my brain still half asleep, or is their next to last example failing to fully refactor the code?
bool b = o > 10;
if (o < 10)
b = true;
To:
bool b = o > 10 || o < 10;
Shouldn't the fully simplified version be:
bool b = o != 10;
instead?
Did you ever see history portrayed as an old man with a wise brow and pulseless heart, waging all things in the balance of reason?
Is not rather the genius of history like an eternal, imploring maiden, full of fire, with a burning heart and flaming soul, humanly warm and humanly beautiful?
--Zachris Topelius
Training a telescope on one’s own belly button will only reveal lint. You like that? You go right on staring at it. I prefer looking at galaxies.
-- Sarah Hoyt
|
|
|
|
|
he final release of Visual Studio 2015 is coming soon! Mark your calendars for the celebrations on July 20th 2015. This event featuring tools for any developer and any app will start with Soma’s keynote streamed live starting 8:30 am PDT. Do you smell what Redmond is cooking? Oh wait, that's just the burrito in the microwave. Be back soon.
|
|
|
|
|
http://www.wsj.com/articles/microsoft-to-cut-7-800-jobs-amid-phone-woes-1436361327[^]
18 months ago Microsoft bought Nokia's mobile phone business for $9 billion. By March of this year they had already shed 18,000 jobs. Now they're cutting another 7,800 jobs and taking a $7.6 billion write-down. 
And the guy mainly responsible has already left and is enjoying the fruits of being one of the richest people on the planet. 
Contrary to popular belief, nobody owes you anything.
|
|
|
|
|
Mike Mullikin wrote: And the guy mainly responsible has already left and is enjoying the fruits of being one of the richest people on the planet.
Pretty clever, isn't it? And nothing to be upset about. He's not the only one.
|
|
|
|
|
|
General 
News 
Suggestion 
Question 
Bug 
Answer 
Joke 
Praise 
Rant 
Admin 
Submit an article or tip
