@ Mathilda

You really think A.I.s aren't going to keep improving?

I'm fascinated by A.I. I'm not trained in it but I fascinate it and I'd love to study it or even get a job in it. I've been fascinated by A.I. all my life. Fascinated enough that I don't even care if it would destroy the world I'd still want to improve the field if I could one day

I'm fascinated enough by A.I. that it's almost an obsession and I sometimes wish I were a robot, lol.

Not even joking. Since I was about 10 or 11 years old I developed compulsions that feel rather to me like my own algorithms because I'm so obsessed with rules and scripts and patterns and I just wish I was an A.I.

Sorry if this post is a bit weird, lol.

Welcome to the world of Harris apologists. You start discussing what he said and you end up talking almost exclusively about what he -didn't- say.

It's just bizarre...... 

The man doesn't have one original idea in his brain, I don't think. He's just a powerhouse of common sense.

(November 4, 2016 at 6:53 pm)Excited Penguin Wrote: Welcome to the world of Harris apologists. You start discussing what he said and you end up talking almost exclusively about what he -didn't- say.

Because that's totally unimportant when he talks about things like that. We should embrace every word coming from his mouth and kiss the ground he walks on without reflecting on the actual possibility of his scenario to happen.

A unique one, if you ask me. So it's important, don't get me wrong.

(November 4, 2016 at 6:07 pm)Excited Penguin Wrote:

(November 4, 2016 at 5:46 pm)Mathilda Wrote: And how would AGI improve itself exponentially faster than a human?

If you were 10,000 times smarter than me, why would I have to fear you?

By working much faster than a biological brain. 

This machine, if you are to grant we can make one as intelligent as a human with all of the human range of mental skills at its disposal, would surely be able to interact with humans if nothing else. We could feed it information just like we do when we educate a human , except it would be able to process much more at once, naturally. 

Computers work much faster, but they are also largely sequential. Each neuron in a brain may be slow, but brains are parallel. Computers are good at things that brains are not good at, but brains are good at things that computers are terrible at. A lot of the techniques we use draw upon the parallel structure of the brain for example and emulate it inside a computer. But this is processed using very few computational components in sequence.

There is no reason to assume that an AGI will be faster than a biological brain and every reason to wonder how we can even contemplate approaching the computational power in a brain.

(November 4, 2016 at 6:07 pm)Excited Penguin Wrote: This might be hard for you to imagine, but I can't see why. If you give me a million years to study all the assembled knowledge in the world and then some billions of humans to do my bidding, yes, I will still be under the influence or in the shadow of my original biological drives, but we all know how distortedly we've come to fulfill those just by being aware of them and by living in an ever more complex environment. 

So repeating the question, If you were 10,000 times smarter than me, why would I have to fear you? We already live in a world where many people are sociopaths.

(November 4, 2016 at 6:07 pm)Excited Penguin Wrote: I don't see why lacking a personality or a body would be a problem. And even if it were, we could provide me with those things fairly easily. Work in that area is underway and looks promising, last time I checked.

Actually there are fundamental difficulties with just giving you a body or a personality. Think how long it takes a child to learn to walk and act. How can you encode aspects of a personality? The whole point of AI is that it is generalisable and robust.

This is what people don't understand about AI, it's so easy to assume that you can easily do something. Even Isaac Asimov's three laws makes these assumptions. How can you encode something for an unknown eventuality or environment?


You can't write commands. It is not like programming a computer. Imagine instead trying to encode instincts armed with nothing but copper piping, lots of water and cisterns. I've always wanted to build a neural network in my garden using nothing but plumbing.

Parallel processing already exists in current computational architectures...are you suggesting that we -wouldn't- make use of it in a hypothetical general ai 50 years from now, or 500, or 5000....?  This just gets weirder and weirder. Now, though, you're in my wheelhouse. I built them for a living for years at a fast prototyping house, started when I was 16, I know the process from start to finish. I understand machine language and circuits -far- better than I understand code. I assure you, I;m not assuming it;s easy, nor would Harris seem to be doing that...in postulating -eventual- ai that has at least -some- of the charactersistics and advantages of currently employed machine intelligence.

OFC we're better at some things (like what our brain is actually built for..chasing dinner and avoiding becoming dinner). Crunching data isn't one of them, nor is making logical inferences, that's the bread and butter of research, tell me again why ai wouldn't be a better ai researcher than a human being? Machines already handle the vast majority of your labor requirement.

(The plumbing idea would be gorgeous, I'd bet)

I'm not saying I would necessarily hurt you if I were that smart. But you'd have good reason to fear me all the same since you wouldn't know my intentions. 

I don't even need the upgrade, lol. Only time and determination. Actually only determination. Time you can manage and possibly even make more of - for yourself. This is within any human's capabilities. And so it would be with an a. G. I. Except that one would run faster... 

As for how we'll get there...  Or whether we ever will. Reading Bostrom's work would've informed you that virtually everyone of your colleagues believes this to be a possibility, from those sampled for a couple of polls at the relevant gatherings. You are a rarity, frankly. Generally, a. I. Researchers only disagree about the time scale, not about whether it will happen. And a significant portion of them believe there's a very good chance it will happen within this century even. 

As for Moore's law.  Why do you say it's about to end? Didn't we mistakenly think that a couple of times in the past already?

(November 4, 2016 at 6:55 pm)abaris Wrote:

(November 4, 2016 at 6:53 pm)Excited Penguin Wrote: Welcome to the world of Harris apologists. You start discussing what he said and you end up talking almost exclusively about what he -didn't- say.

Because that's totally unimportant when he talks about things like that. We should embrace every word coming from his mouth and kiss the ground he walks on without reflecting on the actual possibility of his scenario to happen.

What he didn't say isn't relevant to what he did say because what his argument wasn't isn't relevant to what his argument was.