[evan_au (OP)]

What questions do you think an AI would investigate?
- How would they find out? (if people weren't hand-feeding it)
- What could we learn from a superintelligent AI?
- What do you think?

Yesterday I listened to an interview with James Barrat on the potential benefits and challenges of a superintelligent Artificial Intelligence (AI).
- Such a being would need a crash course on "The meaning of Life, the Universe and Everything", but without the luxury of a 4 to 10 year university course, or even a childhood.
- The first instinct of any living being is self-preservation
- The next priority for a rational being is to assure a future supply of necessary resources
- An AI would need to develop Ethics and Values
- An AI would need to develop common sense about the physical world.
- Like IBM's Watson project, a good starting point may be to download Wikipedia in its entirety (that requires a lot of resources!)

Listen to the interview (1 hour): https://after-on.com/episodes/007

[chris]

The system developed by Demis Hassabis and his team (and bought out by Google) seems to be able to teach itself whatever it likes...

[wolfekeeper]

Values systems are totally necessary- an AI with no value system will just sit there and do nothing at all.

And the problem is the values system. Humans are born with a robust value system that is honed by millions of years of evolution to reproduce and learn, if an AI doesn't have a similar system built into it, one which has to be really rock solid without bugs with unfortunate side effects, then with a super-smart AI that can have unfortunate values then we're potentially going to be in super big trouble.

[Professor Mega-Mind]

 Yo , Diggity ! 
Not alive !  Not a being !  No common sense ! No conscience!
Imitation only !  No soul !
Perfect actor , perfect assassin !  Ya need ta chuck da conceit and grotesque arrogance !  We not be Gods , only men !  Let your mechanical man cut yer fool head off 'cause some pisser wanted a laugh !   .  I don't want "I Robot" anywhere near me .  Ridiculously dangerous , limited technical use only , Mr. I Kill You With My Mistakes !
Alright , 'nuff said . Pull out of the  tailpipe !
   Prof. Meg.
P.S. -
They go from tool , to servant , to  soldier , to guardian , to police , to keeper , to jailer , to sterilizer .  And you are comfortable the whole freakin' time ! 

[evan_au (OP)]

I don't want "I Robot" anywhere near me .  Ridiculously dangerous

You have evidently only seen the movie, which shows robots as being malicious.

The "I, Robot" book (a collection of short stories by Isaac Asimov) is much more tame, suggesting that robots with ethics and values are only accidentally dangerous - and that danger is more likely if people try to weaken ethics and values.

But ethics and values don't sell movies, so the script-writer extracted the most dangerous aspects from a bunch of separate stories, and put them together in a single movie.

Isaac Asimov wrote his "3 laws of robotics" to provide an ethical basis for the robots in his novels. He described these laws as being ambiguous enough to allow for plot twists.

1. A robot may not injure a human being or, through inaction, allow a human being to come to harm.
2. A robot must obey the orders given it by human beings except where such orders would conflict with the First Law.
3. A robot must protect its own existence as long as such protection does not conflict with the First or Second Laws.

A movie with these ethics and values would not sell so well at the box office (for example, Bicentennial Man, from another book by Isaac Asimov).

But we will need a more comprehensive set of values for real robots, hence the work of the "Future of Life Institute".

Meanwhile, militaries around the world are working on AI-based robots, presumably with severely attenuated values and ethics.
See: https://en.wikipedia.org/wiki/Three_Laws_of_Robotics

[Professor Mega-Mind]

         Addendum from P.M.
I read those so long ago I furgot what the dang paperback smelt like !  Asimov's biggest flaw was seeing people as machines ( Foundation anyone ? ) .  None of his characters are incredibly evil or malicious .  In reality , there are hordes of unbelievable fiends staring at computers right now .  They are backed up by wave after wave of demented fools .  " I'm sorry , I didn't know that would happen ! " just doesn't cut it , when you are talking about peoples lives .  To cross story-swords with you , reference 2001 : A Space Odyssey .  As James  Kirk would say " One little thing ! " .  Or as I would say " One rotten bastard ! " .
 I'll save the dangers of replicators for another day .
Make sure you wash that good-looking "car" inside & out after use , y'hear ?
Sleep well and make sure the bot-bugs don't bite !   
P.M.

[guest45734]

What questions do you think an AI would investigate?

Just to keep you straight it wasnt "Al" it was "HAL", which is one step down from "IBM" think about it! H+1, A+1, L +1 = IBM

[Professor Mega-Mind]

...................Replicators
 The thing about replicators , large and small , is their teamwork .  If you combine Moore's Law , great passage of time , and superior , wireless , bot-net capabilities , you get super-machine intelligence , with unimaginable , kamikaze-type , skills and abilities .  The ability to reproduce themselves makes them almost impossible to eradicate .  Indeed , the organic creatures present would likely be eradicated , in fairly short order .
 Prevention is the only safe way to deal with such a profound danger .  Only the strictest laws can suffice .
Vaccine required here .
P.M.

[syhprum]

I have had my life made miserable by a low technology robot that detected me driving at 37 MPH for 2 seconds as I pulled into the traffic stream we need less AI not more

[Professor Mega-Mind]

The potential for people to abuse A.I. is growing by leaps and bounds , isn't it ?
P.M.

[Bored chemist]

I have had my life made miserable by a low technology robot that detected me driving at 37 MPH for 2 seconds as I pulled into the traffic stream we need less AI not more

Does your car have a speedometer?
If it has, then the problem you cite is to do with human intelligence, rather than artificial.

[Galaser]

The potential for people to abuse A.I. is growing by leaps and bounds , isn't it ?
P.M.

Sure,slowly it will become irreplaceable.

[Professor Mega-Mind]

.......Ounce of prevention .
Either humans are riding herd on it , or Reply # 3 happens .

P.M.