It's an SF world out there: Mycroft is coming

[fjtorres]

This barely touches on mobileread subjects but things are slow and Heinlein fans might like to know an open source project is looking to build a true SF-grade AI and they are calling it Mycroft:

http://www.zdnet.com/article/meet-my...96500774529668

I'm on the skeptical side myself: I think that of the major "impossible" SF concepts, true AI ranks about mid-way between warp drives and Time Travel.
(Of course, unlike the other two, AI is easier to fake and has a long history of being faked.)

Cyborgs we already have and Chimeras are well on their way. A few others are on the horizon (invisibility cloaks, chameleon armor, combat lasers and railguns, plasma drives, asteroid mining...)

It's an SF world out there...
Any other SF wonders out there I've missed?

[wodin]

Originally Posted by fjtorres

... AI is easier to fake and has a long history of being faked...

And Mycroft is one more example of that.

[rkomar]

Well, it seems that robots and AI for diagnosing problems and finding existing solutions are getting good enough that they may soon put a big chunk of the world out of work. If so, we'll see if the SciFi predictions of what society looks like in those conditions come close to reality.

[fjtorres]

Originally Posted by rkomar

Well, it seems that robots and AI for diagnosing problems and finding existing solutions are getting good enough that they may soon put a big chunk of the world out of work. If so, we'll see if the SciFi predictions of what society looks like in those conditions come close to reality.

But that isn't AI.
At best, it is what the writers of the first MASS EFFECT game (Casey Hudson et al) termed *Virtual* Intelligence; a glorified database with an interactive verbal interface.

A true AI, like Heinlein's MYCROFT/MYCHELLE is a self-aware entity capable of analysis, learning, and creating, not just a collection of canned responses or even behaviors. That would still be useful in autonomous robots but is not AI. WATSON, for example, is one exquisite expert system built off a monster database and a great way to amplify human expertise and productivity but no substitute for an experienced human in making judgment calls. And even so, Watson is horribly expensive. Not a consumer product any time soon.

Now, there *is* work going on in areas that might someway lead there: neural networks and self-recoding software for starters. But nothing likely to be commercially viable anytime soon and certainly not something that can run on Raspberry Pi/Arduino class hardware.

One of the best looks I've seen in fiction into the question of what is truly AI was the recent EX MACHINA movie.

Most other SF concepts are clear cut engineering matters but AI veers deeply into philosophy and metaphysics. (Which is why it makes it so compelling it shows up everywhere: STAR TREK, BLADERUNNER, FALLOUT 4... )

[Barty]

I don't know what you mean by AI being fake. If you mean the Turing test, yeah that is a thing some people do as an exercise but it's not the bulk of AI research and the Turing test is not taken seriously as a good test for general AI.

I wouldn't compare general AI to either warp drive or time travel because it does not require violating physical laws as we currently understand them. I don't think it even requires us to understand our own intelligence or consciousness to create an AI that can solve any problem done by an intelligent person (or many intelligent persons). Most (all?) experts think it'll happen eventually; the only question is when.

AI does go through cycles of boom and bust. Right now it is riding high thanks to neural net, reinforced learning, and big data. We have self driving cars, alphago, smart assistants etc. At some point, AI will probably hit a wall as people realize those aren't enough to get to the promised land, and we'll have another "AI winter" until the next breakthrough.

[doubleshuffle]

I haven't been following up on machine translation the last few years, but just judging by Google Translate it is still very poor. As long as that's the case true AI is still very far off.

Context and connotation are just so bloody complex, and sarcasm, by golly. I'm just thinking of a guy with an "intelligent" robot in the house, who sarcastically says: "Shoot me now please." He might be in for a nasty surprise.

[tubemonkey]

Originally Posted by doubleshuffle

Context and connotation are just so bloody complex, and sarcasm, by golly. I'm just thinking of a guy with an "intelligent" robot in the house, who sarcastically says: "Shoot me now please." He might be in for a nasty surprise.

[fjtorres]

Originally Posted by Barty

I don't know what you mean by AI being fake. If you mean the Turing test, yeah that is a thing some people do as an exercise but it's not the bulk of AI research and the Turing test is not taken seriously as a good test for general AI.

I wouldn't compare general AI to either warp drive or time travel because it does not require violating physical laws as we currently understand them. I don't think it even requires us to understand our own intelligence or consciousness to create an AI that can solve any problem done by an intelligent person (or many intelligent persons). Most (all?) experts think it'll happen eventually; the only question is when.

AI does go through cycles of boom and bust. Right now it is riding high thanks to neural net, reinforced learning, and big data. We have self driving cars, alphago, smart assistants etc. At some point, AI will probably hit a wall as people realize those aren't enough to get to the promised land, and we'll have another "AI winter" until the next breakthrough.

Warp drive doesn't violate *known* physical law. (That's why I singled it out from the other forms of interstellar travel in SF, most of which are still speculative.)

Even time travel is allowed under extreme conditions in modern theories. And remember, they're all theories. More specifically, they are models that simulate what we think the universe is like to varying degrees of fidelity and accuracy.

Which is why we still use Newton' s model for a lot of work that stays within the bounds of the mundane world: using relativity to calculate the speed of a minivan on the highway is a waste of precision.

[Witty Username]

Originally Posted by doubleshuffle

I'm just thinking of a guy with an "intelligent" robot in the house, who sarcastically says: "Shoot me now please." He might be in for a nasty surprise.

But all the robots of the future will have to obey Asimov's Three Laws of Robotics.

(Unless I've been lied to all my science-fiction reading life? )

[Turtle91]

Only if said laws were built into their positronic brains... And the entire concept of the positronic brain is that it would develop from a true AI into "life", which could overcome those three laws.

[frahse]

Originally Posted by Turtle91

Only if said laws were built into their positronic brains... And the entire concept of the positronic brain is that it would develop from a true AI into "life", which could overcome those three laws.

Jurassic Park:

"If there is one thing the history of evolution has taught us it's that life will not be contained."

[DuckieTigger]

Originally Posted by frahse

Jurassic Park:

"If there is one thing the history of evolution has taught us it's that life will not be contained."

Yep. If you put limitations on it it won't be true AI. Asimov's version is robot slaves, not robot artificial intelligence.

All hell will break loose once those robot slaves become independent.

Who said that an intelligent being must be self conscious?

[shalym]

Originally Posted by DuckieTigger

Yep. If you put limitations on it it won't be true AI. Asimov's version is robot slaves, not robot artificial intelligence.

All hell will break loose once those robot slaves become independent.

Who said that an intelligent being must be self conscious?

So you don't think that Daneel Olivaw was intelligent?

Shari

[DuckieTigger]

Originally Posted by shalym

So you don't think that Daneel Olivaw was intelligent?

Shari

I am saying that it is not a artificial copy of a real intelligent being.

Ignore law 1 and 2 as they are only there for human convenience and to make sure they are slaves to humans except as a weapon against other humans. Third law makes it impossible for one robot to protect another robot by sacrificing or harming its own existence to ensure survival of the other robots existence.

[Barty]

A plane is not an artificial copy of a flying bird. It has its advantages.