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

[fjtorres]

Originally Posted by DuckieTigger

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.

Kinda makes robot altruism hard, doesn't it?

No, Asimov robots aren't AI and were never intended to be AI.

(Adam Link was. https://en.m.wikipedia.org/wiki/Adam_Link)

Asimov's robots were always intended (within lore) to be appliances (mostly because he wrote the first few to debunk the "Frankenstein Principle") but as he kept exploring the material they really became an exploration of human foibles (even cold analytical Susan Calvin ended up projecting her needs on one) and the Law of Unintended consequences. Complex systems created by fallible humans will themselves be fallible in unpredictable ways. (Notice how Skynet always screws up? )

R. Daneel was not intended to be sentient but in the end, just like Mycroft, sentience emerged and he ended up substituting his judgment and agenda for humanity's. He claimed it was in service to protecting humans...

But, lets face it: we hear that all the time from busybodies, politicians, and other villain types.
(Asimov was a great writer but his politics and personal philosophy were a tad Pollyanna-ish.)

We don't need to go too far to find and example of human fallibility in trying to fake AI:

http://money.cnn.com/2016/03/25/tech...tay/index.html

Imagine the same problem popping up in a true AI, which would be millions of times more complex. Just like a child, you can only control what they learn while in closed environments, and sometimes not even that. Not saying the Frankenstein principle is right but all evidence to date is that a true AI will be at least as unpredictable as humans and almost certainly more so. No biological imperatives at work: we have an eon or two of genetically ingrained responses that make us semi-predictable, especially on a statistical basis--Asimov himself saw this. His PsychoHistory is almost certainly possible. Another SF concept headed our way.

And one to beware of, at that.

[frahse]

Mistake...

[fjtorres]

With the recent hype and hoopla over "robot personhood" CNET has a couple of mostly rational looks at AI.
First, the fictional depictions:

http://www.cnet.com/news/hollywood-a...t-far-fetched/

(And yes, Clarke's explanation in the books of how Hal "went bad" is pretty Asimovian. For a non-Azimov construct. Hal was just following orders. Conflicting orders.)

http://www.cnet.com/news/ai-frankens...e-experts-say/

For now, AI is good at sorting things, making predictions and powering through algorithms, but it struggles to think critically or solve problems you'd find in standardized tests. The Allen Institute for Artificial Intelligence recently devised a contest in which the computer program that scored the best on an eighth-grade science test would win $50,000 for its creator. They all failed. The best score was a 54 percent.

In other words "AI" is good at data processing and lousy at actually thinking.

It'll stay in fiction for a while longer.

[capidamonte]

While Mycroft is billed as AI, and is certainly in that wheelhouse, I think a better way to think of it is as an vocal user interface (VUI?). So the skills mentioned here (sorting, predicting, etc.) are exactly the sort of thing you need to achieve the end result of talking to your computer for those tasks in which talking is the best and easiest interface.

Think of how swiping and pinching are generally an excellent way to manipulate images onscreen. It can be done with mouse and keyboard, but it is definitely not as easy. Voice command is going to work for things like media playback, information searches, smart-home control, and probably a million other things. It's going to be great for people with physical disabilities that make computer-use difficult, too.

Mycroft is also intended to be a locally controlled system unencumbered by the security and privacy and reliability concerns caused by corporately-owned advertising and data-mining interfaces disguised as utilitarian services. Think about Nest, the "smart home" control system that shut down its services recently, stranding people who bought their hardware with no options but to throw their hardware investments away. Any service controlled on other people's computers is primarily subject to their interests, not yours. And the people who control things like that never turn out, in the long term, to be the people that created it in all their wide-eyed naivete. You certainly will prefer all your Internet-of-Things devices to be under your own local network's control if you think about it for even a couple of minutes.

There is still a piece of Mycroft (voice-to-text) that is being run on the company's servers, but that is changing even as I write this. The creators are already committed to an open-source, self-hosted option.

Another great thing about Mycroft is that it can eventually be federated to all your chosen devices so that you will have what is basically the equivalent of your own, personal Jarvis. And no one is going to use your habits, thoughts and dreams to market to you or manipulate your choices and opinions.

If you're not technical enough to run it on a server in your home or business, there will be paid providers available -- guaranteed. And their privacy and security models will also be open to scrutiny. The safest code is open code and Mycroft is going to be a great option for those who treasure their privacy and dignity.

So even if AI never gains sentience (something I'm not sure I'd be comfortable with anyway, both on an ethical and practical level) something like Mycroft is the answer to some of our awkward interface issues, and our dangerous societal issues.

[Greg Anos]

If you're willing to brave the Politics and Religion forum, I had a thread on some of the"fringes" of science.

https://www.mobileread.com/forums/sho...d.php?t=266923

After all, what is science fiction but playing games with the fringes of science?

[crich70]

The movie Ex Machina
https://www.youtube.com/watch?v=bggUmgeMCdc

explores the topic of A.I.

[grumbles]

Artificial intelligence will never be a match for natural stupidity.