The Singularity Institute is a research organisation that’s as forward-thinking as most Gizmodo readers (read: sci-fi nerds).
Give $US1
They recognise that, as transistors double on a microchip every two years, and as artificial intelligence takes over our daily life (including our public transit, stock markets and every single Google search you do), that maybe this idea of computers morphing into our new robot overlords is more than a cheesy-but-lovable sci-fi flick.
With eight full-time staff and a budget of under $US1 million annually, they’re mathematicians and programmers attempting to find a way to develop AI that will be “altruistic” rather than indifferent towards humans, or worse yet, malevolent.
They’re re-evaluating the very way we conceptualised AI from the very beginning – research that actually began in the ’50s – in hopes of creating a new system from the ground up, if necessary. It’s a unenviable task without a sure solution in sight.
How do you program something that will eventually have god-like intelligence to think a certain way when that day comes? Truthfully, it may be impossible, but the Singularity Institute has a bit more faith in technology than their fellow man.
“Human selfishness is a part of evolution to pass along our genes,” explains the Singularity Institute’s Michael Anissimov. “That wouldn’t necessarily follow with another type of system we built to be altriustic or benevolent from scratch.”
If you’d like to support the Singularity Institute, you can head over to Philanthroper today (disclaimer: I totally run that site) and give $US1 to their cause. I take nothing from that $US1, and 99 cents makes it directly to Singularity Institute, which is a better rate than through their own site.
Don’t feel forced to donate. Just keep in mind, if you don’t, you’ll be the first guy who “volunteers” to run a vest full of dynamite through the front door of Skynet. [Philanthroper and Singularity Institute]
Timmy Toucan
Friday, April 1, 2011 at 9:21 AMHmm. The main problem with donating would be: What if they fail?
I don’t want our new robotic overlords to know I was working against them.
Josh
Friday, April 1, 2011 at 9:55 AM+1
olearymo
Friday, April 1, 2011 at 10:13 AMThe thing is, they key to it would be giving them a sense of gratitude, that’ we’re their creators. But the thing is, this would require something probably best described as ‘love’.
Would it be possible to make them feel this, if any emotions at all? I’d say it’s as unlikely as machines gaining sentience at all.
As in, very unlikely. I don’t believe it’s possible for a machine to become truly self aware, no matter how complex it is.
That doesn’t mean, however, it couldn’t be dangerous. Machines are tools, but tools ALWAYS do exactly what they’re told. If we slip up and accidentally allow programming in something that could turn dangerous… only ourselves to blame, it wouldn’t be the machine’s fault.
Really, I feel that in the narrative of Terminator, Skynet never was *actually* self-aware or sentient. It was just doing something that was programmed (some interpretation of ‘protect yourself, and strive for perfection’, possibly).
Matt Larritt
Friday, April 1, 2011 at 12:37 PMYou can defienantly program love, it’s essentially the knowledge that you need that person/thing for whatever reason and they make you feel primarily good things. Love is essentially good feelings and a bit of personal logic, nothing more. It’s chemicals released through your body which are the same chemicals released from experience with that person… When you’re thinking about that person, or being with them. These chemicals are released whenever you look at someone, it’s basically combined emotions of what that person means to you. If you hate them, they release anger… If you love them, you have less angry memories, and the good ones trump. So effectively, all you need is a receptor saying “This is good, I don’t want this to be harmed” and signals being released whenever this loved one is thought about/noticed… What you need to do tho, is hard program this in, so the robot has no ability to change their emotions regarding any particular person…. Essentially, it is a hard-wired response to people, and shouldn’t be allowed to be modified at any time.
Think of it like this, in our brain, we score people on our experience with them. We don’t score in numbers tho (Although, should be possible in programming), we score in chemical response. So what happens is, if you’re walking down the street, and you meet a complete stranger, and the stranger smacks you in the balls, that action will release a chemical response in your brain/body which signal this person is pissing you off. This is stored in the brain as a memory, so next time you see this person, your memory contains that chemical response, so it will again release it and you’ll know this person has pissed you off, so this confirms you don’t like this person… If this person then buys you a present, apologizes and gives u a good reason to forgive, then suddenly the feelings may change about this person (Depending on your logic) so this person recieves a new coctail of emotions which changes the release of chemicals when this person is recalled in memory… So essentially, if you love someone, it’s basically that they’re useful, they release mostly good emotions when you recall them from memory, and you feel you need them in your life.
All we have to do, is make sure robot’s have the inability to change the way they feel about humans and we be sweet…
I can talk about this for ages… I feel like it’s possible to port the human brain into a computer, you could probably even run it in a simulated environment… Hells yea, this is where heaven actually exists, without a body, you can’t die. In a virtual space you can exist for as long as you want and do whatever you want (limitations to the computer/developer of environment depends)!!!! You could even be uploaded into the T-1000 and go for a stroll in the real world…
Ross Weekes
Friday, April 1, 2011 at 4:58 PMSorry Matt, But I doubt this. You see what any program lacks also is creativity..
At the base of all programming is this:
*A program, no matter how complex, is still a bunch of programmed responses.
A human being is not this. A human being can for instance question its existence. A program cannot, unless you write in the question (In which case it’s just another programmed response isn’t it!)
msgbox(“What is the meaning of my existence?”)
wscript.echo “Follow the white rabbit…”
Matt Larritt
Friday, April 1, 2011 at 5:28 PMSorry Ross, but you assume we are able to ask questions where there are no questions. The thing is, everything we can imagine is a collaboration of our own personal experiences. If you were born, strapped down unable to move, completely seperated from the rest of the world, in complete silence fed thru tubes sort of thing… 15-20 years later I’m pretty sure your creativity won’t exist and I’m pretty sure you won’t have any questions to ask. What is exposed after this is bare animal behaviour, you will be able to study what a person is at the core, without life interferring. This is how our mind works, our questions are completely based around our individual experience in life. Social structures are able to modify the way people grow to be, this could be true for computers also. If AI is programmed to be able to question it’s own existence without knowing the root of the question, what difference would it have between it asking itself the question, and you asking yourself the question?
Robot’s with the ability to learn and modify their behavior through other robots/people, hence they will believe they exist, they believe they exist because we program them to believe they exist. It’s all about how the robot percieves things. Pain – You could punch a robot in the face, and have it programmed not to like that experience (Give it a negative emotion value of say, 30%), so the robot would not want to go through that again. Next time you try and punch it, it’ll probably move away from the punch, just like you would… Kick it in the nuts, it would give it a negative value of 40%, so the robot would rather be punched in the face then be kicked in the balls… It didn’t know this however until it experienced the symtom. We feel pain to indicate damage to our body… There is absolutely no difference to how we percieve things, and how a robot can be programmed to percieve things, even it’s own existence.