Hacker News Reader

Comment by wcarss | Hacker News Reader

Described in this way, it feels like a reasonable argument that humanity is like a paperclip-optimizing AI that just maybe just chooses random things to go deep on optimizing.

We created a problem for ourselves that was fun -- games -- but once we'd beaten "games", the reward function kept running and now we're adversarially inventing new ways to "optimally" beat harder and harder versions of "games", ie new objectives, new limitations, new allowable orderings of elements within the games, etc.

I wonder what 50 years of this will yield!

View on HN

I was thinking of it like a mating display. Supposedly the goal of such a display is for the male animal to demonstrate its superior genes by wasting resources in a show of wealth, so it must purposely seek out something that is difficult or expensive to accomplish.

Heh. It's pretty amazing how much of life boils down to this.

I think the greatest form of it may be the act of saying "fuck you!" to every bell curve imaginable. Fuck bell curves, I'm going to do something else.

I really want to start a cult of bell curve hate.

Welcome to the other side! :-) Life is good here.

>> it feels like a reasonable argument that humanity is like a paperclip-optimizing AI

> I was thinking of it like a mating display.

What is "life", if not a paperclip-optimising AI applied to our own DNA?

I conjecture that you are right and the reason this works is that a large class of optimizing goals actually converge in the long run.

In other words, whether you choose paper clips, Bibles or human DNA as your optimization goal, over very long timescales (like a billion years) the optimal algorithms for all of them converge. If you want to make a solar mass of paper clips, you need a lot of infrastructure in common with a solar mass of any other goal.

There does seem to be a kind of turnpike theory-like effect to technological development.

I'm reminded of Schmidhuber's PowerPlay (https://arxiv.org/abs/1112.5309), too. It just trains on increasingly difficult problems with no direct objective function beyond solving the current problem and devising a harder one.

that argument definitely has some truth, but it's worth considering that playing a randomizer is a very different experience than speed running the original game as mxwsn said. I've played a few randomizers and it's a whole lot of fun to dynamically route based on evolving probabilities - vs performing a known route at a hyperoptimized level. randomizers reward adaptation, as well as gameplay and game knowledge that may not matter for the speed run

Reply on news.ycombinator.com