Technological Singularity / Computer Science

[Float]

How might a technological singularity affect the supply, demand, education and work of computer scientists?

 

[Brontosaurie]

it's a very interesting subject which i find gets complex and unpredictable rather soon, making speculation difficult.

wouldn't computer scientists be crucial and in high demand during a transitory phase (which of course we've entered), and then either redundant or norm?

 

[Architect]

By definition we don't know what will happen post singularity - of which "nothing" is an answer - but I think that programmers will always be needed. In fact in the new world order we'll probably be the only ones receiving the gratitude of our new masters.

 

[Brontosaurie]

By definition we don't know what will happen post singularity - of which "nothing" is an answer - but I think that programmers will always be needed. In fact in the new world order we'll probably be the only ones receiving the gratitude of our new masters.

to be fair, he didn't state that the question pertained exclusively to post-singularity. the approach ought to be relevant too, no?

 

[John_Mann]

By definition we don't know what will happen post singularity - of which "nothing" is an answer - but I think that programmers will always be needed. In fact in the new world order we'll probably be the only ones receiving the gratitude of our new masters.

I think jobs and economy will be no more before the singularity. With 3D printing from molecular level, why do you need to buy something? What about slave robots (imagine Asimo 100x more evolved)?

I don't expect some partial gratitude from a super intelligent being. If the gratitude is a logical thing our overlords will be friendly to everyone. If someone it's not so good in something we can't know for sure it was an intentional decision or it's for a lack of opportunity or even capacity.

I think intelligence brings benevolence. Just think about the human geniuses from now and past, benevolence is a trait of them.

After all, I don't know for sure if a non biological machine with super human intelligence is a possibility. But is a fact we can enhance our intelligence through technology, maybe the future super intelligence will come from enhanced humans.

With or without singularity technology itself will collapse the world as we know it (very soon, in 2 or 3 decades).

 

[John_Mann]

to be fair, he didn't state that the question pertained exclusively to post-singularity. the approach ought to be relevant too, no?

Oh yeah. We can wonder about if just the concept itself already affects computing area today.

I think so.

 

[Analyzer]

I think the biggest question regarding singularity and the role of automotive tools is the economic question. Are we going to enhance an even more centralized coercive society or keep it decentralized and voluntary as it naturally begins?

 

[Cognisant]

I predict that sometime relatively soon a form of artificial intelligence will be developed that can understand human concepts and abstractions, this will be taught/programmed to understand computer science, it will become the perfect tool. We won't need to know how to program or how our computers work because our technology will understand itself, I'll be able to say to my computer "I want a program that can convert wireframe models into a 3D printable format" and the AI will know what I mean, it'll design and develop the program for me faster than any human programmer could.

From there on it's hard to predict but possibly computer science itself could become automated and soon thereafter as far as we're concerned the distinction between technology and magic becomes moot. Which is not to say people won't be doing sciencey stuff, y'know if making shit happen was as easy as waving a stylus in the air while reading from a thesaurus of scientific terminology we would be doing it constantly.

 

[John_Mann]

I think the biggest question regarding singularity and the role of automotive tools is the economic question. Are we going to enhance an even more centralized coercive society or keep it decentralized and voluntary as it naturally begins?

Exactly. What the world will be like with no more scarcity? I think the more close we can get is the concept of anarchism. For the first time we'll be able to experience what is individualism. One virtual "tabula rasa" universe for every individual*. In this reality what could drive us to meet with another "real" individuals?

* We don't need real people in these personal universes, we can just make realistic simulated people by using philosophical zombies, just like in games.

 

[John_Mann]

I predict that sometime relatively soon a form of artificial intelligence will be developed that can understand human concepts and abstractions, this will be taught/programmed to understand computer science, it will become the perfect tool. We won't need to know how to program or how our computers work because our technology will understand itself, I'll be able to say to my computer "I want a program that can convert wireframe models into a 3D printable format" and the AI will know what I mean, it'll design and develop the program for me faster than any human programmer could.

IBM Watson is running this way. He(?) already understand natural language and abstract associations.

 

[Cognisant]

No it dosen't.

Watson is a fancy search engine, like google but focused specifically on trivia and using a wide range of pre-programmed techniques to deduce what is and is not likely to be the answer, if you want to stump it first ask one question then follow up with a second question: why?

 

[John_Mann]

No it dosen't.

Watson is a fancy search engine, like google but focused specifically on trivia and using a wide range of pre-programmed techniques to deduce what is and is not likely to be the answer, if you want to stump it first ask one question then follow up with a second question: why?

Watson it's not only a trivia search engine and it's pre-programmed techniques involve random outputs that's impossible to predict.

Of course Watson doesn't know "why" but that's another step. But, as I said, Watson already understand natural language and abstract associations.

Surely Watson it's a great step towards AI. Watson is a direct ancestor of the future AI.

http://www.youtube.com/watch?v=DywO4zksfXw

 

[Cognisant]

It's a publicity stunt by a behind-the-times computer company.

Of course Watson doesn't know "why" but that's another step. But, as I said, Watson already understand natural language and abstract associations.

The ability to identify words is a program I can buy today for $50 (or likely get for free from sourceforge) and there's really only so many ways you can ask a question, so yes Watson can deduce what the question is even when it's being asked in a odd way but that's really far from being "natural language", as far as I know Watson dosen't consider the tone of voice in which the question is asked or which words are emphasised, both of which are factors that can completely change the meaning of the question. Furthermore Watson can't deal with emergent meaning (such as if I make up a word in order to simplify my explanation of a complex subject) or slang, or even sarcasm, so to say Watson can understand "natural language" is a flat out lie.

Likewise though Watson technically can do associative mapping, what it does is really statistical analysis whereas what our neural nets do... there isn't really a word for it, suffice to say it's not just statistical analysis, our neural nets are combined recognition and reasoning engines whereas the statistical analysis Watson does is neither, it's just number crunching.

If you want to see something that impresses me look up the quadrupedal robot that learns how to move from scratch just from input & output, sophisticated learning like that is the future of AI, Watson is just a very well built expert system, y'know it's an impressive achievement in of itself but it's not really doing anything that hasn't already been done elsewhere, it's an amalgamation of yesterday's breakthroughs, befitting a behind-the-times company.

 

[Architect]

It's a publicity stunt by a behind-the-times computer company ...

Wrong on both counts actually. IBM is doing cutting edge work as evidenced by Watson.

The ability to identify words is a program I can buy today for $50 (or likely get for free really statistical analysis whereas what our neural nets do... there isn't really a word for it, suffice to say it's not just statistical analysis, our neural nets are combined recognition and reasoning engines whereas the statistical analysis Watson does is neither, it's just number crunching..

To the best of our knowledge that isn't true. Watson works quite similarly to our brains, which look like Hierarchical Hodden Markov Model engines. Obviously Watson lacks the spark of true intelligence, but in other ways (instant access to vast knowledge stores) it is better.

Read How to Create a Mind by Kurzweil to understand these issues and the state of research.

 

[Analyzer]

If you want to see something that impresses me look up the quadrupedal robot that learns how to move from scratch just from input & output, sophisticated learning like that is the future of AI, Watson is just a very well built expert system, y'know it's an impressive achievement in of itself but it's not really doing anything that hasn't already been done elsewhere, it's an amalgamation of yesterday's breakthroughs, befitting a behind-the-times company.

Yeah I see your point in that IBM is just trying to maintain or be relevant in today's world. It appeals to baby boomers or the older generation who are familiar with IBM and its role in the computer industry. If it was any other new company it wouldn't get the same marketing and exposure, probably because they don't have the same finances. Still Watson is impressive in itself but the approach seems very SJish.

If what your saying about the future of computers and AI, then there will be a transitional phase into it. What do you think are key things for humans to actually advance past that phase. Or rather, what key problems need to be solved in the meanwhile?

 

[Cognisant]

Edit: In response to Architect.

I don't care how effective an expert system it is the fact of the matter is that it's not learning things on it's own terms, which is why as far as I'm concerned it's just a fancy program on a fancy computer, whereas the low level not-particularly-effective stuff impresses me because when it does get developed to Watson level sophistication the results are going to be far more impressive.

Watson isn't cutting edge, it's just expensive.

 

[Analyzer]

What's wrong with sourceforge?

 

[Cognisant]

I'm a diehard believer that if you're truly serious about making advanced AI your primary focus should be adaptability and this is because the less you design a system, the more it's designed to design itself, the further that system can be developed, so the ideal would be a program that could be applied to any task, let me repeat that again for emphasis: ANY TASK.

Once you have that you have a system that's nearly without limit, if it has to learn everything from scratch then it can learn anything from scratch, y'know I can say now that Watson will never be self aware, it just won't happen, the program just is not that adaptable, it can have trivia on itself but without a foundation of knowledge upon which to contextualise that trivia it won't understand, it will speak but it won't know what it's talking about, whereas a program that can, that is designed to figure things out itself for itself will have the potential to develop itself to that level, it could become self aware.

More importantly intensively adaptable programs could learn from us, learn to understand us, and once man and machine can establish a two way dialogue, once we can understand each other just as a easily as you can understand me now...

I cannot explain, it would be like explaining the global network of airports and flying machines to someone from the dark ages, how do you convey such things?

 

[Cognisant]

What's wrong with sourceforge?

Absolutely nothing

 

[bartoli]

Bumping this thread, this made me think about the singularity :

The Linux kernel is at the heart of every GNU/Linux distribution, Android and any number of embedding devices. Linux is virtually everywhere, running desktops, laptops, servers, routers and phones. The kernel is made up of millions of lines of code (version 3.10 if the Linux kernel contains nearly 17 million lines of code) and thousands of changes are included in each new release. This raises the question of who writes all of this code and The Linux Foundation has some answers. According to their report, Red Hat contributes the lion's share of kernel code, with approximately 10% of changes coming from Red Hat employees. Intel, Texas Instruments, IBM, Google and Oracle are also big contributors. Linux developer Greg Kroah-Hartman made some comments with regards to the pace of Linux development, observing: "Rate of change, and number of patches accepted, keeps going up as our community size has increased," indicating the pace of development is continually growing. "Every year, since 2003 when I started keeping these numbers, I have said, 'Wow, we are going so fast, there's no way we can keep rate of change and number of developers and companies going,'" Kroah-Hartman said. "And, every year, that number has increased. So I'm a bad judge of these things."

(source : http://distrowatch.com/weekly.php?issue=20130923)

 

[RaBind]

Listening to some of Ray Kurzweil's talks gives me the idea that in the singularity/post-singularity robots won't take over, but in fact we will become the robots, to transcend from our biological limitations.