Do you think AI (artificial intelligence) is something to be feared?

Sorry man it is not personal but your opinion is stupid and short sighted as unfortunately you will learn later on...now tell me how shall I put that in more diplomatic wording without being an hypocrite and confusing the crowd?

PS - ...oh youtube videos are the best way to learn if you know were to go to...it seams you have some trouble with it...yet another unfortunate sentence coming out of your mouth...how shall I address that? This continuing stream of wrong thinking? The best I can do is vent off at stupidity. Its done for millennia did you know?

Again with the personal attacks. I guess I just need to accept you for who you are. You learn from Youtube videos. I learn by doing.

Machine learning is my job. I am a software engineer; my job is working on a expert system with the goal of interacting with humans to provide useful information in human languages. This is what I get paid for, and my team is quite successful. I happen to know what I am talking about.

You can attack all you want based on what you glean from the internet. You should know that the people who are actually working AI don't get information from YouTube. We study, we go to school, we write software, we go to conferences, we write and read papers. We run experiments. We get results. And we bring useful products to market.

I have knowledge and real world experience on this topic. I can't make you respect that. But I wish that you would be open to listening to another point of view (no matter what you think of the individual) without having to resort to childish name calling.

I respect your right to hold to your opinions.

Most engineers working in AI (in my experience) think that Elon Musk's ideas about AI are crazy. Bill Gates agrees that Elon Musk is crazy.

Unless you start saying something meaningful stick your authority in the field where the sun does not shine! You are that kind of "I point you what to do" Engineer and have zero understanding forth sight on just how dangerous AI exponential development is gonna be...You probably one of those fools who believes we are on the verge of a 2 Industrial revolution...your analysis on how these systems have patterns and their own rhythms that unwrap our very own way of living is leaving you blind. On topic, unsupervised learning is already being developed as we speak and goes straight to the heart of the matter that I pointed you out in the first posts...at some point as the need for human management decreases and the capacity for these algorithms to learn increases AI will be developing software to itself, to its own needs, rhythms, in its own pace...we, as far as I can see either follow up with brain implants as Elon Musk suggests or are bound to become zoo pets living quality life without a say in anything! Of course this is to far out for you am I right?

Just see the fracking video that is not 4 minutes long...I always care enough to post a couple for the average Joe and a couple more for those who want to put the time in!

Unless you start saying something meaningful stick your authority in the field where the sun does not shine! You are that kind of "I point you what to do" Engineer and have zero understanding forth sight on just how dangerous AI exponential development is going to be...You probably one of those fools who believes we are on the verge of a 2 Industrial revolution...your lack of analysis on how these systems have patterns and their own rhythms that unwrap our very own way of living is leaving you blind. On topic, unsupervised learning is already being developed as we speak and goes straight to the heart of the matter that I pointed you out in the first posts...at some point as the need for human management decreases and the capacity for these algorithms to learn increases AI will be developing software to itself, to its own needs, rhythms, in its own pace...we, as far as I can see either follow up with brain implants as Elon Musk suggests or are bound to become zoo pets living quality life without a say in anything! Of course this is to far out for you am I right?

You get all of this from watching YouTube videos....

You are making the claim that someone who actually designs and maintains machine learning systems is "blinded" by this experience, where as someone with no experience is somehow free to derive truth from Google searches and TED talks. It is a really good thing that you didn't study machine learning in college Fil... or you might be unable to pull your expertise out of the air like this.

That's really kind of funny.

I don't mind you having opinions, and you are free to pull them out of anywhere you choose. It is the personal attacks (against someone who actually has first hand experience with machine learning systems) that are quite ridiculous. I claim no magic ability to see the future... and sure, there is a possibility that some dangerous AI will develop in the near future. I would be willing to discuss this possibility. But my opinion is that this is very unlikely, and I base my opinion on real experience in the field, rather than YouTube videos.

If you could state your opinions without the personal attacks... it would be nice.

Man are you thick...
Your continuous appeal to authority just shows how insecure you are about anything.
Since when in the entire history of science working in the field was guarantee of quality or understanding eh?
The paradigm has been exactly the opposite...
And just what is that you have against online learning you dinosaur? You are pathetic!
There, a couple of more ad hominem attacks hard earned and well deserved!
Finally and to the point if you have an argument rather then blank authority make the argument on why there is no problem with the development of AI...as far as I can remember in the thread other then reassuring without making a case you haven't justified much. 60% of the world population is about to get unemployed and you think its fine because Hal 3000 is yet to come up. Smart!

This might be an interesting topic for another thread Fil. And if you would like I will start it.

The great revolutionary Scientists; Galileo, Newton, Darwin, Pasteur, Einstein... they all worked hard to master the current scientific understanding before they went to change it. They all studied, read and worked to master their field. These aren't people just pulling ideas out of the air. These were people who worked very hard to master a field before revolutionizing it. They were educated, they learned as much math as was available. They had access to the top expertise of the day.

You need to have expertise in a field in order to understand it, or to advance it. These ideas don't come magically to uneducated people. And sorry, you don't gain any level of expertise or understanding through watching YouTube videos.

Expertise is important. Experience with a field, hard study, experimentation, access to the expert community... these are all things that are critically important to understanding and advancement in any field.

We can start another thread if you disagree with this.

I am sorry for this tangent with Fil. I would like to get back to the topic.

Is there anyone here who is interested in a deeper discussion of how machine learning systems actually work? If so, I would be happy to engage with a perspective based on current experience working on these systems.

Thank you for making my case all of those names you mentioned above often worked against the status quo in the field, exactly the kind of blind authority you are appealing to right now! Do you have a solid well rounded argument against mass unemployment or not? Bring it!
You degree of denseness is such that you lack the basic of most basic grasp that the implications of AI development have nothing to do with being able to program in AI. The consequences are sociological economical and political and require an holistic analysis on how we should deal with it! Are you an expert on those areas to?

On sociology analysis it is quite enlightening to see a group of scientists think the current problems in the world, terrorism, mass shootings, and dystopia online, are all caused by religion...that says a lot on the capacity of missing the point even when we refer to the supposed best of best around...

For the sake of example the guys who invented the web bluntly state that not in their wildest dreams they had the vaguest idea what they were conjuring when they made it...

Yes, Al Gore was always a little clueless.

Am I right in saying the argument so far is:

A: I work in AI, and it's no where near being able to self learn - everything type of learning it can achieve is through directed programming

B: AI is dangerous. Most of the worlds jobs are going to be lost. Just look at these youtube videos.

A: They are scaremongering. Informed perspectives say that's not currently possible.

B. The youtube videos are informed perspectives! Didn't you watch them?

A. Yes I did, and I work in the field. Which is why I can tell you that they are scaremongering

B. Just cause you work in the field doesn't mean you know Jack Shite.

A. True - I can't say that I know who Jack is - he's probably a lovely lad... but getting back to the topic - in order to understand if something is clap trap or legit, you need to have done some real study.

B. Poppycock.

A. draw up an organisational chart to show how you would program AI. Then write enough code to show you know how, or don't need study...

I added the last bit in. I'm presuming that's where this is going.

Two other notes:
- I don't see how anyone who hasn't studied programming AND AI could ever make the leap to developing true AI. Certainly from that point of view, a person saying 'it cant happen' has weight to their opinion; and
- it might be possible. Eventually, once you teach computers to start writing their own code (I've heard, but not confirmed that this is starting to happen) - this is something our brain does <neurons connecting>, and the ability to learn <and I've heard there's advancements in this> is also something our brain does...currently such 'behaviour' still fall within programmed parameters. Certainly many advancements have come from people going against the grain, sheer creativity, or moments of genius.

I agree with you here, or at least I think I am agreeing with what you are saying here.

I do a little programming myself. I am extremely skeptical about the claims where the "AI" systems google has, have been "writing" their own code. Especially when they claim even their "coders" are completely baffled when they look it over and can't make heads or tails what the code does.

To me there is some bullshitting going on here. What programmer would not be so ecstatic to not want to delve in and get to the point they can understand the code? What I mean is, it can't be so foreign that it's impossible for a coder to look at it and have ideas what the code does. You can't have infinite possibilities with the code. There are still limitations, rules. You can't simply break the rules.

I honestly think its people who are not coders, who are completely clueless about writing code that are making the public statements about google AI writing it's own code. And that it's foreign to their coders. I think this is a PR attempt to bullshit the public to get them interested in a narrative.

I would be more than happy to be completely wrong. But I highly doubt that I am wrong about this.

Yep. Programmers should still be able to look at code and go 'ah hah, I know what it does'. Programming still uses a language. It's like teaching a computer to speak in English, then claiming 'it said something in English, but I don't know what'

That said, all language has structure, and that structure can be taught (or in this case, programmed). And each word/function has meaning. It should be possible for a program to write simple code by
- programming in an understanding of the language (though why you would want to write the incredibly complex code to allow a computer to understand its programming language enough to write simple code, is beyond me...just program if x then y).
- programming in context (itself so incredibly complex)
- with the next problem would be checking the results of that code. Not only would you need to program a way to simulate it the outcome (which is really what any computer generated code should do first), but you would need a way to understand the outcome, and so many outcomes are subjective.

Now if it were me, then I'd start looking at a 'reading humans' program, where they tell how a human reacts to it's decisions by reading the chemical, heat, breathing, facial changes...to determine subjective outcomes <but you see how complex this would be, seeing as we don't fully understand body language ourselves>...and in any case, once a person figured out how it determined subjectiveness, a whole heap of people would try to manipulate the computer.

But really, if it were even possible for a persons mind to be that organised, detailed, and creative...it still looks like more effort than it's worth.