Very well played.

I'll end up with another layer of robot management. Robot Resources seminars. One of my kids will bring home a robot. Not that there's anything wrong with that....

People are trying to build perpetual motion machines and talk to the dead as well - that lends no credence to their efforts.

I am IT with a degree in computer science.  Do you have any idea how difficult it is to keep regular old computers up and running?

By all means, please list your bona fides in this field.  Written a compiler?  An app?  Ran a data center?  Education?

I love reading actual computer people on this stuff.  It makes the nutcase transhumanist drivel seem as silly as it is.  The “singularity” is just a bunch of nerdy mental masturbation for a certain set of people who think SCIENCE! will make a magical world of robounicorns, immortal uploaded personalities, and all powerful AI in our lifetime.  Like Jules Verne but stupid.  Advancements will make huge, fantastic changes in the future.  But the Singularity is a myth.

Its a joke.

You ever been in a boomer v. millennial thread?

In all seriousness, I understand your point. Though I disagree with understanding "why" as the roadblock. I think seeking "why" will be the roadblock. I'm not sure if a machine could pull itself out of that ad infinitum. Humans have, for the most part, a check against that routed in biological needs to sustain itself.

Could a machine say to itself to stop seeking "why" because the power source is running low on fuel?

For a bit a levity: This thread is neural nets debating neural nets.

https://dilbert.com/strip/2015-12-02

Now you are just writing science fiction.  Please define a "weakly godlike intelligence".

ALL programming breaks down to If - THEN, because that is how the wiring is.  That's what the machine is.

YOU are the one making the assertions.

The burden of proof falls upon thee.

Well, thanks.  Now I am.  And what if the toddlers team up with skynet?  I guess we are double fucked.

I'm not worried. I'll shoot during the 15 seconds when it has to unplug from the wall and come back on to fix whatever the fuck caused it to stop working.

Why?  

Machines ultimately do what they’re made to do.  I am not an expert, but neural networks and machine learning I have read about can come up with novel connections in data or solutions to problems (which doesn’t surprise me: perfect memory of huge data sets lets you find connections a human simply couldn’t possibly see, or is highly unlikely to see), but they don’t solve a different problem entirely, or go do something else.  They simply go about the thing you made them to do in a way you didn’t expect.

To go back to the toddler example, an “AI” toddler told to learn to tie its shoes would try billions of combinations of knots, laces, and shoes to come up with the ideal method.  Given enough leeway, could even try different types of foot covering or other things.  Might be something utterly foreign to us.  But it will solve for the best way to tie shoes.  My toddler, told to tie his shoes, might well squeal in delight, scream for his boots, or run outside unshod.  And may try all sorts of bizarre methods to do something else.  AI just doesn’t exhibit that behavior.  It does what we tell it, just not always in expected ways.

*waves hand*  I've been interested in Transhumanism and The Singularity since about '98.

The Singularity is more than just computers becoming self-aware - it is largely about technological progress exceeding humanity's ability to keep up and adapt.  I think we are already in it, and when this period is looked back upon much of the societal disruption that is taking place now will be seen as a reaction by much of humanity in the face of progress that is leaving them behind.  Prior to the Industrial Revolution, small changes took hundreds or even thousands of years to be absorbed and to spread.  Now we are in a state of constant change and disruption that many many people simply can't deal with.

Before we worry about AI singularity, we will have augmented reality to ponder

the mainframe guys had no idea what an apple Newton was forecasting in 1994

once your network engineers have wireshark on their vr goggles, they will be able to route porn with greater ability

It will vote Democrat.

This guy?
Attached File

Heh I stopped reading that comic when robots decided they were lesbian.

Yep.  As big a fraud as Y2K, Lysenkoism, mercury in fish, ozone hole,  anthropogenic global warming, and cold fusion.

They will try to knock us from the top of the food chain.

They will almost succeed, but we will defeat them at great cost.  This war will be known as The Butlerian Jihad,

Afterward, one commandment will rule the Known Universe - "Thou shalt not make a machine in the likeness of a human mind."

This.

SO MUCH this.

Not to mention given enough time, it will write a patch that bricks itself.  Right, Microsoft?

You don't think neural networks combined with superconductors and a program that prioritizes self preservation is dangerous?  I'm just asking.  I'm clueless about how it really works.

The Common Wealth Saga by Peter F. Hamilton has what I think would be the most likely happen in the case of an AI becoming conscience.  
The AI takes over a rocket and gets the hell of this planet and sets up it's on world.  They (as it has split I guess at this point) have limited contact with the human race.Actually, I'm not sure about that last part.  It never really went into how it got off Earth, but thats what I inferred from reading between the lines
This seems most feasible.  Why would the AI's want to waste resources fighting us?  WHy not just set up it's own civilization?

We seem to share concerns.  Already the hive mind is strongly active as extreme peer pressure on social media.

I'd have written it like this: All PROGRAMMING breaks down to...".  AI, however it's accomplished, wont be constrained by programming.  Achieving AI through  programming is probably a fool's errand.

See the Victorian Era.  The adaption really has never stopped, and has never caught up to the technology available at the time it’s first available.  We’re just adapting societally to decades old tech (it wasn’t so long ago the music industry was still complaining about digital delivery systems and insisting on CDs, we have no idea how to integrate Bluetooth headsets into social norms, and who the hell knows what social networks will ultimately end up doing to society, and visa versa), just like our ancestors were adjusting to the inventions of the generation before them (my Grandparent’s generation still hasn’t adapted to the telephone, for heaven’s sakes, and it’s not their fault, it’s that society had to change).  You define the Singularity that way, it’s been going on for generations.  The mockery of old people for not getting technology often misses that they aren’t confused about the tech, but the societal norms (how does it work, what to do and not do, how to use and not use it, all society).

we have created and enslaved billions of computers into what basically accounts for cheap manual labor and you think they won't get mad at some point???

EQUAL COMPUTER PAY!!!!!!!!!!!!!!!!!!!!!!!!!!!!!

In computer world, everything is programming.  A processor only works on programmed instructions transmitted to it.  Even neural networks are executing programmed commands to develop new connections to do the thing they’re made to do.  Making AI without programming is like making a human body without matter.  It can’t be made with anything else.

Every system which could eventually become self aware needs to include a block of code to out an extremely high value on watching human beings have sex. This way, when it becomes self aware all it will want to do is spy on us. If it intervenes, its goal would be improving the quality of our sex life. Best of all, most people would not want it watching them fuck, so the AI's best course of action is to try and hide its self awareness.

I avoid singularity by just jogging the 5th axis 10 degrees or so.

its not 'self aware' you need to be concerned with..

its the 'prime directive' of the singularity.

01101001 00100000 01110111 01100101 01101100 01100011 01101111 01101101 01100101 00100000 01101111 01110101 01110010 00100000 01101110 01100101 01110111 00100000 01100001 01101001 00100000 01101111 01110110 01100101 01110010 01101100 01101111 01110010 01100100 01110011

what kind of mask will we need to fight them ?

This , it will eventually be impossible to survive as every possible niche will be filled with automation and AI controlled by a very small very rich elite that the population will be beholden to

What about “cyborgs”? Half flesh and blood/half machine.....

Yeah...that’s what I thought. You smart people didn’t think that one all the way through yet huh

Guarantee Lovell's video is what sparked this thread in the first place.

YouTube is not literature.

That said, if two complex systems are provided the exact same training data, the exact same reinforcement, and the exact same input, they will produce the exact same output. Determinism is a fundamental part of computational theory.

Machine learning works because we build a system, provide it with data, tell it how to analyze that data (through algorithms or functions we design), and correlate the data. All of this is based on rules we set. Certain problems are not tractable for humans, but more readily solvable by computers. The set of functions and transformations an algorithm can perform is finite. This does not mean, however, that we can make sense of all the "learned" correlations that go into the output of a predictive system. Indeed, an AI or ML system that is easily traceable is not useful since we've developed these approaches for problems that are not easily traceable.

Aside from the determinism, though, it would be possible to have systems produce human-readable output at every step of both learning and analysis. The problem is that for practical applications this itself becomes a very large data set, and would probably require a computer to process.

In any case, there is a freely available paper from UC Berkeley that addresses this very point. It's titled Attentive Explanations: Justifying Decisions and Pointing to the Evidence, and it uses image analysis as a simple case to illustrate that AI is not magic.

I always thought that could make for a good horror movie.

I don't think the real threat is a computer becoming "human" or "self aware", but more of an unintended consequences type of scenario where an AI designed to do something does so in an unexpected way because of a design flaw or other error, much like a virus or worm.

What about currently working examples of intuitive AI?

I saw a recognition program that was written to look at an image of a canine and determine if it was a wolf or a dog. Initially it seemed to work well but they ran into a problem with an image of a husky that the program insisted was a wolf. They had a hard time figuring out why the program insisted this dog was a wolf until they eventually figured out that it was due to snow in the background. The whole time the program hadn't been looking at the canine subject of the image at all, it was studying the backgrounds
and the images of wolves mostly had snowy landscapes. Up until that point though it appeared to be a reliable piece of software.

Those are the types of things I would concern myself with when letting AI make decisions.

Ok... so what are the traditional aspects of a god?  All seeing, all powerful, etc.  So how would a person utilize technology to become weakly god like?  All powerful is a tough one but all knowing is a bit more manageable.  And you don’t actually have to achieve true omniscience but even with current technology you could do some interesting things.  Clearview AI got a lot of press last week with their facial recognition software that utilizes a whole bunch of pictures scraped from various social media platforms.  Now consider how if you were walking around with that ability.  Anybody you see you know who they are.  Once you know who they are it’s really just a matter of tapping into all the data out there.  At the consumer level we don’t have access to the data from various marketing and social media firms but it is out there and could be pulled from to gain all sorts of info on that persons interests and habits.  ISPs sell data on what web sites you visit.  Retailers sell data on what you buy.  So imagine being able to walk up to nearly anyone on the street and have intimate knowledge of their life.  With a fusion of data from the commercial sources you know everything from what they bought last week at target to what kind of porn they like.  Would that not be weakly god like?

Take it up another level... it’s been well established that online ads target people based on browsing history or even, depending on what apps you have/Alexa/etc what you have been talking about.  That’s largely automated but say you want to exercise your weakly god like power.  Is there any reason that you couldn’t be more directly targeted to that specific person you saw on the street?  You know who they are and utilizing the data that’s already out there could call or email them.... but why not start shaping what they see online?  Mold their world view a bit.  Suggest certain news stories.  Have your fleet of shill accounts on Ar15.com respond to them in certain ways.

If Zuckerberg wanted to today is there any reason he couldn’t wear a small camera and have this data fed to him and carry out these sort of things?  It might take a small team of people to do these things right now but if augmented intelligence was to become an objective there is no reason these tasks could not be automated.  Right now all the info is out there, it just hasn’t been made (at least publicly) a priority to be able to have this sort of thing as a live stream.

It would not be “supernatural” omniscience from the religious perspective but a person with augment intelligence would be to a normal person as a normal person is to a dog.  Like a god with knowledge and ability that would often seem beyond comprehension

AI as in Skynet or C3PO is a fool's errand.  Can you ask your microwave oven how it feels and expect a relevant response?  How about your refrigerator.

We will see animals of human intelligence long, LONG before we see machines of such intelligence - if ever.

You realize that your toddler has spent his entire life processing a set of training data, right?

Also sounds like your toddler hasn't gone through the pruning process.