Print

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

[Graknorke]

Aww don't be such a stick in the mud, they're cute little wheel robots.
Cute!

[Frumple]

Nooooo, no nuking. We need that sort of program. That's how we make major strides in material science, engineering, and all sorts of other things. Autonomous iterative improvements come up with stuff humans are terribly unlikely to chance upon, and can bring pretty impressive gains in certain fields. Think it was some kinda' wicked effective antenna that serves as a good example of the results?

There's going to be a lot more coming in the near future, though, as we're either at or near breakthroughs on the programming side of that sort of project. Which is good! Massively good. The stuff that's going to come from it is likely to advance some of our sciences by years... possibly decades or more.

[Akura]

@Truean's second link:

You know what? I searched for that movie on Youtube, and I am absolutely pissed off that Youtube put up a link for that movie which is actually just them saying the video was removed and that you should buy it from Youtube, followed by a full 2 hours of a black screen.

[TempAcc]

I'm all for self replicating advancing robots as long we keep a leash on things. The "machines will be an core part of humanity's future" thing isn't an exclusively futurist argument anymore. The common agreement is that, if true, self improving AIs are possible, we'll probably create one before we reach the end of the century, and we'll either be replaced by it or integrate ourselves with it.

Both outcomes are acceptable to me, but I prefer the latter one :v

[Frumple]

*waggles hand* It's not really a self-replicating robot. It's a builder bot set to try to build better wheel bots, more or less, apparently with iterative/evolutionary design principles. Sorta' like replacing an engineering team trying to implement a six sigma type plan for a specific product with a specialized production program. As the article notes, it can (and similar projects have in the past) come up with interesting designs a human specialist (or specialist group) just wouldn't think of.

There was actually a simulated form of this that's running online posted here in GD a lil'bit back. Forget what it's called, something about letting vehicle designs "breed" to produce better machines. Thing in the article is roughly the same concept, just taken offline, heh. Give it another few decades and we'll probably be seeing a lot of our general use stuff (cars, buildings, etc., etc., etc.) designed largely the same way.

[Rolan7]

Spiral energy - the unstoppable force of an sexually dimorphic race.
These programs aren't stored with DNA helixes, but combining designs and determining the fitness of the results is ooooold hat   Oldstyle genetic modification.
As Psycho Mantis said, "This cow makes a lot of milk so have it [procreate with] a lot of other cows"

[Arcvasti]

There was actually a simulated form of this that's running online posted here in GD a lil'bit back. Forget what it's called, something about letting vehicle designs "breed" to produce better machines.

Do you mean boxcar2d? That's what first popped into mind when you mentioned that.

[EnigmaticHat]

*waggles hand* It's not really a self-replicating robot. It's a builder bot set to try to build better wheel bots, more or less, apparently with iterative/evolutionary design principles. Sorta' like replacing an engineering team trying to implement a six sigma type plan for a specific product with a specialized production program. As the article notes, it can (and similar projects have in the past) come up with interesting designs a human specialist (or specialist group) just wouldn't think of.

There was actually a simulated form of this that's running online posted here in GD a lil'bit back. Forget what it's called, something about letting vehicle designs "breed" to produce better machines. Thing in the article is roughly the same concept, just taken offline, heh. Give it another few decades and we'll probably be seeing a lot of our general use stuff (cars, buildings, etc., etc., etc.) designed largely the same way.

So what you're saying is, the research team implemented a plan to make research teams obsolete

[Eric Blank]

Perhaps this research team was specifically tasked with researching more effective research team structure/administration.

I'd expect its more of a "we're too lazy/unimaginative and there are too fucking many possibilities to explore each one extensively on our own, with our current budget, so we're going to run this program that gives us some neat ideas to focus our attention on instead of finding all the dead ends the old fashioned way" deal.

[Rolan7]

Research team is developing a system where they get to play more DF.

[Frumple]

So what you're saying is, the research team implemented a plan to make research teams obsolete

Not quite, but it's along the path of that.

Which is, like, ideal. Humans are pretty bad at a lot of things, but we're good enough to make tools that are better at it, whatever it is. Or at least make tools that can make tools (that can make tools, etc.) that are.

Though yeah, as per EB, even something that can "just" (scare quotes because there's nothing merely satisfactory about that) winnow out the dead end designs in a comparatively short time frame is basically a gorram coup for a design team. And everyone else in their field. And eventually most of the human species, as the design/research teams waste less and less time on pointless junk and more time on breaking the physical world and its optimal forms over their knee. Efficiency, not laziness. Or both. Motivation don't matter much so long as it's getting results.

[TheBiggerFish]

Hit 1,500 emails from colleges today.
AAAAAAAAAAAAAAAAAAAAAAAUUUUUUUUUGGGGGHHHHHHHHHTOOMUCHSTUFF

[EnigmaticHat]

Relevant CCP Grey video.

[Lord Shonus]

Relevant CCP Grey video.

Hate to tell you, but that video is full of shit. The "job killing robots" that he goes on and on about would cost thousands of times as much as human workers.

[Graknorke]

I have managed to write a thousand character in 3 hours of sort of trying. I'm pretty sure that I could easily write more than that on bay12 in a day.