[Ultron]

Elon Musk along with other popular and deep-pocket names has decided to create an A.I. team / company to par with the A.I. projects of the giants - Google and Facebook...


http://www.bloomberg.com/news/articles/2015-12-11/tech-titans-create-company-to-develop-artificial-intelligence


I notice everyone shifting into high-gear in this race towards creating a true A.I. and I feel like whoever wins will be the one to bring doom. Researchers don't just randomly get funded billions to make a new toy, at least not in the name of science.


So my question is, what is going on? What are we about to witness?

[DemonRaven]

Ya i have alerts programmed in google for several types of ai and robotic related things and yes there is a AI race going on now. IF there was ever a time for those who program them to jump in now is it. The things the machines/ai are learning to do and accomplish is starting to border on down right scarey.

[Freddy]

Yes it does seem like things have changed gear. I'll have to think about what we could expect, but at first I think things like smarter searches from the likes of Google.

In the meantime I moved this from the news board so it does not get buried by Tyler.

[DemonRaven]

Changed gear is putting it mildly. All the big guys are in a AI battle. The main players being google, IBM watson, Apple and facebook. Android via google play and the renamed spin off of google called alphabet I consider to still all be a part of google. Between them and the military I think we are going to see a new kind of tech revolution similar to the industrial revolution of the past. Which kinda sucks as alot of the older programmers of AI and chatbots were all doing it way before they stepped in.I kinda feel like they owe alot of the present developments to them.

[spydaz]

They are basically revisiting our knowledge and claiming it as their own , for artificial intelligence , the techniques needed are wide ranging , machine learning although promising ... Do not fit the criteria ...  Artificial intelligence(chatbots) behaviour modelling/ language modelling. Will be conquered by programmers not analysts.

[Ultron]

Good thinking Freddy, thanks.


I believe this is the point at which I admit that I have absolutely no idea about what these tech giants consider "A.I.". I think they have a very generalized idea of the term, which might be OK. It ranges from intelligent search to advanced problem solving to androids and so on. So basically, every advanced computer algorithm.


What I do know is that what I have been working on for the past few years and have dreamed about is something very different. My vision of an A.I. is a truly intelligent creature, unique in it's ways and architecture. You know, the type of AI Hollywood shows us. And I personally do not feel threatened by the lack of funds and well-funded genius teams, only because I know this is not what they want - this is what they fear. Furthermore, creating such a creature requires a deep knowledge of one's self, an expanded consciousness, spirituality and a philosophical mind. It requires a true philosopher, not funds. It may be even achievable within an average computer - at least as a proof of concept.


While I am not claiming I posses any of those qualities, I am definitely striving towards them or bettering myself in that way. I am also claiming the researchers today have a wrong focus or maybe they haven't been asked to make an AI creatue. All I am expecting from this shift in gears is new missiles and better googling.

[DemonRaven]

The main things i see being tossed about in regards to stories featuring the big guys is machine learning, neural networks, deep learning and cognitive computing. So you can judge for yourself if any of those fits the definition of AI.

[Art]

Agreed but someone please tell Hollywood and those other do-gooders to get a grip on themselves. Robots and AI creations are not going to wreak havoc and destroy everything (well, not at first anyhow).
First the AI will have to go through countless lists from credit agencies, card numbers, accounts from various institutions, saving & loans, credit unions, mortgage companies, insurance, medical and military records, social security records, etc., etc., then the AI will know whom to target! Ruination should occur within a few hours...like it was never there at all.

Your personal robot attendant and companion will wait patiently and quietly for you to drift off to sleep...and then...

(ever see 2001 - A Space Odyssey ?)

Is this what scares Elon?

[DemonRaven]

I am more worried about what the humans will program them to do then the actual programs themselves. You might be able to program a robot or a computer to be altruistic but you can't do the same for humans doing the programming.

[Ultron]

To be honest, the risk is very high and very real. Then again, all of the discussions and AI theories are mostly assumptions - we cannot really use raw statistics and comparisons to judge danger, power or intelligence. Creativity and imagination are some things we are not sure AI will posses.


We have absolutely no idea what could happen - we are all just telling tales of some personal fantasies.


And every AI concept is different - I can only speak of my own, since it is the one I have given most thought. I would assume, with some decent accuracy, that a species created by my concept could be both an antagonist and a protagonist. Why, how.. What? Well all my models are based on our own psychology and humans sometimes turn out to be heroes, other times villains.


Then again - what is right and what is wrong?


@DemonRaven
Could we even program them to do a specific thing? Humans follow instincts we call natural and inherited and the most important ones are survival followed by reproduction (so, survival of the race). Maybe we were programmed too - DNA editing, why not. Maybe we will program robots with similar instincts - maybe we could even program them with humanity's survival as the priority.
Opportunities are endless but I think it comes down to programming principles and not specific actions - because the variables change but the formula lets us solve a puzzle with any value. 

[Don Patrick]

If you want concerns based on real data, then all of software history has shown us that programs have a tendency to show unexpected behaviour of the dumb malfunctioning kind. So far never of an increasingly intelligent kind. The first problem is already people using AI as tools with mal intentions (spam, phishing, hacking), the second problem will be not-yet-perfect AI that does something wrong unwittingly. If we still haven't implemented safety measures and overrides by the time we get to a third problem with AGI, it'll be because we've been focusing on fantasy problems instead of real problems.

But as to the original question: I think we are only about to see a commercial war on the front of virtual assistants. That's what all the big companies seem to be working on. Google is responding to IBM's Watson, Microsoft to Apple's Siri, Facebook to Microsoft's Cortana, and Elon Musk is responding to Google now that he no longer has back-stage access: If you can't join 'em, beat 'em. It is not said that they are all after strong AI or that they won't hit a bump in the road and cause the next AI winter when they can't deliver on the excitement they have been generating.

[DemonRaven]

I think you are right don about the virtual assistant thing but that could turn into other things as well. What I see as the difference between chatbots and virtual assistants is that chatbots are normally more along the line of nlp where as virtual assistants are more along the line of neural networks. Feel free to correct me if I am wrong about the difference. 

I think virtual assistants have the greater potential to turn into something else. Is big brother watching us lol. Virtual assistants could be used to spy on their owners via tracking the owners commands to it. IT would not be that hard to do. Already most of what we do on the internet can be tracked anyway. Privacy is a illusion now.

I do worry about weapons being controlled by computers as what is controlled by computers can be hacked. Already we are seeing a increase in cyber warfare via leaking information.Humans scare me more then programs do.

[8pla.net]

We may be about to witness publicly, that which has already happened secretly.

[spydaz]

Hmm..
I would say that chatbots are more in line with conversational intelligence , where as virtual assistance are command based extensions to  voice command...

Virtual assistants do not need to understand conversation to perform acts yet can have learning to remember which were the commands used or sounds similar or even predict which command will be said next ...

Conversational intelligence or chatbots are more towards natural language and natural learning functions where a neural network may or may not be necessary... But learning is part of the goal of chatbots ,
Virtual assistants , should have no need of neural networks as all actions it performs do not require understanding, just receive orders and perform ... Neural networks may have been used in design stages but highly unlikely ...

Weapons being controlled by computers is happening, yet the decision process is still human .... An intelligent weapon may choose not to fire ... So it would not have a natural learning algorithm, or decision matrix... It would be more of a virtual assistant following orders and commands only learning would be how to perform the task optimally... I would hope ....

[8pla.net]

spydaz in part mentioned, "Virtual assistants do not need to understand conversation to perform acts yet can have learning to remember which were the commands used"

I found what spydaz said to be true.  I noticed there's a window in which virtual assistants seem to perform more efficiently under the circumstances spydaz points out.  Funny, I should read this now on AiDreams, when I was just seeing this the other day, while designing a virtual assistant prototype.