Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[Helgoland]

A progressive's obituary for one of the leading figures in anti-abortion campaigning. "Credit where credit is due"

[Baffler]

A progressive's obituary for one of the leading figures in anti-abortion campaigning. "Credit where credit is due"

I was expecting something shrill and crude¹, but that was actually kinda nice. I certainly learned something from it.

¹the comments did not disappoint.

[alexandertnt]

The alternate solution to the murdering AI is just, you know, not plug guns and bombs into it.

We don't even need a general AI, all we need is AI which is only capable of making decisions within it's programmed context. This AI is all around us now, Google use AI extensively in their search engines, for example.

Keep in mind that almost all the jobs robots would replace don't require any particularly advanced AI, most of them are repetitive menial work with only basic decision making involved.

[Putnam]

Yeah, AI in a box and all that. It could perfectly convince you that letting it into the world at large would benefit humanity more than anything else you could possibly do and that not doing so makes you more evil than any other human in history.

[ArKFallen]

Yeah, AI in a box and all that. It could perfectly convince you that letting it into the world at large would benefit humanity more than anything else you could possibly do and that not doing so makes you more evil than any other human in history.

And the worst part is it could be right. Depending on if you define evil through intent, actions, or results. The latter two would be the clinchers.

[ChairmanPoo]

I've always found the "AI in a box" experiment to be pretty silly. People don't operate on RPG-dialogue-tree logic.

[Dutchling]

I've always found the "AI in a box" experiment to be pretty silly. People don't operate on RPG-dialogue-tree logic.

I like shorts! They're comfy and easy to wear!

[ChairmanPoo]

name!  job!   bye!

PD
.. you know, I think I'm going to try communicating in Ultima7ese for a while. Just to see how well it works out...

[Descan]

The alternate solution to the murdering AI is just, you know, not plug guns and bombs into it.

We don't even need a general AI, all we need is AI which is only capable of making decisions within it's programmed context. This AI is all around us now, Google use AI extensively in their search engines, for example.

Keep in mind that almost all the jobs robots would replace don't require any particularly advanced AI, most of them are repetitive menial work with only basic decision making involved.

Er... No. All the jobs a robot would replace would be... All the jobs. Programming, research, busyness management, anything a human can do a robot or AI can do, conceivably. It's already happening.

Why would you think they'd stop? I don't even know how to describe that concept.

[Frumple]

Yeah, it's one of those things that are kinda' important to remember when talking about future developments in automata -- while humans are pretty good at a lot of things in a subjective sense (relative to other "natural" entities), we're really kinda' terrible at most everything in something approaching an objective sense (from the perspective of maximal efficiency or effectiveness). There is basically nothing a human can do that a machine will not, eventually, be able to do better, and, in all likelihood, cheaper (though that usually comes a fair amount of time after the better side of the equation).

And that holds true for pretty much everything else. Evolution is kind of a terrible engineer -- really good at what it does, but really bad at actually achieving goals beyond that particular focus, solving problems, etc. Machines -- AI, expert systems, simple automated systems, whatever -- are almost certainly going to eventually replace... well, basically everything. Because they're better at it (it being whatever task or tasks are being considered) than we can be. Since they're, y'know, specifically built for it, whereas more or less nothing else is. Barring some kind of catastrophe or specifically sabotaging ourselves in that arena, the question of humans being effectively replaced in all roles is not a question of if, but when.

... and we kinda' want that to happen, imo. Means we can goof off, do whatever, and still avoid starving to death in the streets. Probably. It's certainly a greater likelihood of success than pretty much any other scenario for eliminating human suffering. Chance of robot apocalypse, sure, but at least in the scenario of a robot apocalypse there's (probably) a legacy left behind, as opposed to all the other sorts.

[Bauglir]

And keep in mind the possibility that, soon after the AI achieves superintelligence, it may figure out how to help us become superintelligent, too.

[Leafsnail]

I've always found the "AI in a box" experiment to be pretty silly. People don't operate on RPG-dialogue-tree logic.

It's literally just based on a Terminator movie, and not even one of the good ones.  Personally I'm more worried about zombies.

[Antsan]

And keep in mind the possibility that, soon after the AI achieves superintelligence, it may figure out how to help us become superintelligent, too.

Or the super-intelligent AI might be developed attached to our NI from the get go.

[ChairmanPoo]

And keep in mind the possibility that, soon after the AI achieves superintelligence, it may figure out how to help us become superintelligent, too.

Or the super-intelligent AI might be developed attached to our NI from the get go.

NI

[alexandertnt]

Er... No. All the jobs a robot would replace would be... All the jobs. Programming, research, busyness management, anything a human can do a robot or AI can do, conceivably. It's already happening.

Many of those jobs listed are ones that people enjoy doing, there may not even be a need to replace them with AI if people voluntarily do them. Humans may not even be intelligent enough to create an AI smart enough to handle some of these problems beyond what a human can do.

And regardless, the original point still stands. Build an AI specifically for Programming, for example. It doesn't need to know how to interact with people or how to use bombs, it just needs to churn out computer code. AI research is very little like what you see in the media, it is focused on producing AI for very specific roles, too many AI Winters have been caused by over-optimistic researchers claiming the creation of more general AI will happen any day now. A self-driving car is incapable of composing classical music, Iamus can't drive a car, etc.

Why would you think they'd stop? I don't even know how to describe that concept.

Because there is no longer an impetus? If robots take up enough roles that people are more free to choose what they want to do, and the output is enough to meet demand, why would you build a robot to do your job? Why would a programmer who enjoys programming want an AI to do their job?

There is another interesting point. Perhaps Humans are not intelligent enough to create an AI that is itself capable itself of creating AI more intelligent than itself. At that point, AI becomes not very useful (beyond the certain specific fields that AI do are better than humans).