Print

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

[LuuBluum]

I can assure you, whatever complexity you seem to think exists with DF's personality system, does not really exist.

[MaxTheFox]

"there are other things to be solved out there, that are much more discovery of the century than machines that know how to teach themselves."

Machines can already teach themselves, see neural networks. But even the most sophisticated cases aren't alive or sentient.

It would take the kind of mind that can see a whole universe dancing around in field interactions for to understand, and DF can help with that

How?

It CAN solve strong AI, with minor tweaks from what it is.

HOW? We're listening.

[Jarhyn]

It CAN solve strong AI, with minor tweaks from what it is.

HOW? We're listening.

Asking me to tell you how to create strong AI when it's one of my own life goals to do strong AI first is kinda self-defeating if I actually want to publish, ya?

As for helping someone see what kind of field interactions define a whole functional universe, it's kinda facile: if you can see the whole thing, not for sprites and graphics and "dwarves" and "spaces" but as (data) and (process) instead, it helps someone see all THIS not for the narrative elements but for the (data) and (process) of it all. That doesn't free you from "that which may be narrated", but at least it helps you understand what story it is that's being told.

Really, it's the simplifying transform offered by a much less complicated universe in a jar you can "look down at" rather than one you have to "look up at".

It's a lot harder to understand the concept of "universe" when you can only understand it from "looking up".

[MaxTheFox]

Asking me to tell you how to create strong AI when it's one of my own life goals to do strong AI first is kinda self-defeating if I actually want to publish, ya?

Good luck I suppose. But DF's personality system is too simplistic to even be close to strong AI. Look elsewhere.

As for helping someone see what kind of field interactions define a whole functional universe, it's kinda facile: if you can see the whole thing, not for sprites and graphics and "dwarves" and "spaces" but as (data) and (process) instead, it helps someone see all THIS not for the narrative elements but for the (data) and (process) of it all. That doesn't free you from "that which may be narrated", but at least it helps you understand what story it is that's being told.

How is that helpful? DF is too different from the actual universe as its rules are very situational and vary wildly depending on how you interact with it. Our universe, from what we can see, is governed by universal rules. DF is not fields and particles. NJW put it best.

Really, it's the simplifying transform offered by a much less complicated universe in a jar you can "look down at" rather than one you have to "look up at".

It's a lot harder to understand the concept of "universe" when you can only understand it from "looking up".

But how is that useful to creating strong AI, especially since this "universe" (in quotes because it only is one metaphorically) isn't even internally consistent and its inhabitants don't actually learn anything? I literally do cellular automata as a hobby. Those are closer to being universes in the normal meaning due to being internally consistent and emergent, yet they are simply too "low-level" to do anything even resembling AI with. Try again in 50-100 years I suppose.

[LuuBluum]

If your life goal is to create strong-intelligent AI and it's entirely contingent on having access to the Dwarf Fortress source, I suggest rethinking your goals. Or at least, putting more effort into actually doing this than just pondering what it would take, and actually sitting down and writing code.

You can probably pull off the personality system of Dwarf Fortress as a standalone thing in a few years, depending on how much time a day you dedicate to it. Time spent waiting for something that's never going to happen (having access to the DF source) is time wasted not accomplishing your goal.

[Magmacube_tr]

i pley funi migdet gaem

[Putnam]

It CAN solve strong AI, with minor tweaks from what it is.

HOW? We're listening.

Asking me to tell you how to create strong AI when it's one of my own life goals to do strong AI first is kinda self-defeating if I actually want to publish, ya?

i would not trust anyone who says this with a text transformer of sufficient eloquence, much less strong AI

[Laterigrade]

Not only am I highly skeptical of the ability of a game to prove Riemann's hypothesis, what on God's green Earth is "Altering the Carbon"? At least explain how you would go around doing that.

Well, if you whack it a couple times, I’m sure it’ll end up altered, one way or another.