Games discussion

[AI]

Except humans CAN & DO learn. It's not an esoteric concept. It's very common and very simple.

For humans, that is.

For computer's it is impossible. The best an AI can do is draw on the parameters input into it. Processing anything outside the scope of it's reality is incomprehensible. In fact, the comparison of human and code is so overly simplistic that it is difficult to even contrast them. The AI's only world is the game. It plays the way it is told to play. Even the best AI's have hardly any capability to react; rather, they act. Humans can take their experience from real life, from other games, from their friends, and apply those things to the game.

Compare a computer to a human, fine. But comparing a human to a computer is more than insulting. It's misleading.

That's a very limited view of 'computers' (this is really more about algorithms). Learning algorithms are quite possible and are used in a number of fields. Usually though, you can strip out the learning overhead by pre-running the algorithm until it's well-tuned and then releasing that.
The problem with AI in many games is problem size (you can't check every possible move) and heuristics (assessing a situation *without* calculating every option). One heuristic that works well in deterministic games is "checking which option worked best the last time", so, simply try an option, and if it doesn't work, don't do it again. I'd call that learning.

Essentially, what humans are good at is heuristics: pattern matching. Taking lots of shortcuts to arrive at a reasonably good result quickly. It does have it's flaws though, if it didn't, things like optical illusions wouldn't exist.

[Insinuator]

That's a very limited view of 'computers' (this is really more about algorithms). Learning algorithms are quite possible and are used in a number of fields. Usually though, you can strip out the learning overhead by pre-running the algorithm until it's well-tuned and then releasing that.
The problem with AI in many games is problem size (you can't check every possible move) and heuristics (assessing a situation *without* calculating every option). One heuristic that works well in deterministic games is "checking which option worked best the last time", so, simply try an option, and if it doesn't work, don't do it again. I'd call that learning.

Even the best algorithms still operate in an extremely narrow field. They can only see what parameters are allowed to be available to them. Humans can reach outside of what is "allowed" and create something entirely new. The gap between AI heuristical "reasoning" and human heuristical analysis is in quantity and quality. And it is a monstrously overwhelming maw of a gap.

The only reason AI's can beat humans is because they operate in narrow, confined universes of logic. This is evidenced by a game's complexity. Generally (for not all games are created equal), the simpler a game, the harder AI it has. The converse is true as well. More complex games generally have dumber AIs. This is not just a lack of skill on the programmer's part. It is the amount of data the AI must process and the increasing number of exploits the human player can realize. Think Pong vs Wesnoth. I don't think there is a player out there who could beat Pong at it's maximum potential. Yet no serious Wesnoth player would argue that the AI (without bonuses) is a serious challenge.

[aklotz]

Except humans CAN & DO learn. It's not an esoteric concept. It's very common and very simple.

For humans, that is.

For computer's it is impossible. The best an AI can do is draw on the parameters input into it. Processing anything outside the scope of it's reality is incomprehensible. In fact, the comparison of human and code is so overly simplistic that it is difficult to even contrast them. The AI's only world is the game. It plays the way it is told to play. Even the best AI's have hardly any capability to react; rather, they act. Humans can take their experience from real life, from other games, from their friends, and apply those things to the game.

Compare a computer to a human, fine. But comparing a human to a computer is more than insulting. It's misleading.

That's a very limited view of 'computers' (this is really more about algorithms). Learning algorithms are quite possible and are used in a number of fields. Usually though, you can strip out the learning overhead by pre-running the algorithm until it's well-tuned and then releasing that.
The problem with AI in many games is problem size (you can't check every possible move) and heuristics (assessing a situation *without* calculating every option). One heuristic that works well in deterministic games is "checking which option worked best the last time", so, simply try an option, and if it doesn't work, don't do it again. I'd call that learning.

Essentially, what humans are good at is heuristics: pattern matching. Taking lots of shortcuts to arrive at a reasonably good result quickly. It does have it's flaws though, if it didn't, things like optical illusions wouldn't exist.

What I see an algorithm as doing is basically filtering out anything that does not act like the algorithm. Essentially an algorithm raises the bar for HUMAN learning in the sense that the human MUST fulfill the goal of the algorithm to compete with it and essentially can only match, never win. Isnt that thinking flawed by a programmer whose own view of the world may be flawed by the fact the standard must be as narrow and bounded as input and output. If you sit at a computer all day and to communicate with it must become more like a computer to gain the ability to program it isnt your world view contained quite massively to something a computer could probably duplicate quite easily? Look at a standardized test as being an algorithm. Isnt that what it is? Remove anything that has obligations outside and above the purview of the test material? Build cities of people who sit at computers that train other people to do the same thing that is already known. What a world.

In gaming the AI just cheats flat out to keep the illusion of a true challenge. Sort of like, hey if you can match at this level you have clearly sat at a computer far too long. It seems ego driven more than anything. More like ok, we've got nothing more but lets just lie about it.

[Insinuator]

Spoiler:

In gaming the AI just cheats flat out to keep the illusion of a true challenge. Sort of like, hey if you can match at this level you have clearly sat at a computer far too long. It seems ego driven more than anything. More like ok, we've got nothing more but lets just lie about it.

Hey, hey, now. Ease up on the programmers. Programming a decent AI is a ton of work, and the people who do it should be respected for their diligent efforts to create even the facisimile of intelligence. I think most AI coders have good motives in trying to create challenging, "creative" algorithms in their AIs, not just beating other people. It's just that there are limits to what they can create, so they must substitute more obvious boosts to the AIs power.

I don't want to derail this topic, so I have nothing more to say.

[aklotz]

Spoiler:

In gaming the AI just cheats flat out to keep the illusion of a true challenge. Sort of like, hey if you can match at this level you have clearly sat at a computer far too long. It seems ego driven more than anything. More like ok, we've got nothing more but lets just lie about it.

Hey, hey, now. Ease up on the programmers. Programming a decent AI is a ton of work, and the people who do it should be respected for their diligent efforts to create even the facisimile of intelligence. I think most AI coders have good motives in trying to create challenging, "creative" algorithms in their AIs, not just beating other people. It's just that there are limits to what they can create, so they must substitute more obvious boosts to the AIs power.

I don't want to derail this topic, so I have nothing more to say.

Yep there were some cheap shots for sure. Sorry about that.. People are def working hard and for good things. I absolutely enjoyed my time 'gaming' and pre-multiplayer, AI was really essential to play. I guess now botting has become more of a nuisance but i do value the intent and know a hell of a lot of work went in so dont take criticism the wrong way. I am being narrow to get a certain point across about my own frustrations with 'gaming' today as i've gotten older, NOT to belittle the hard working folk out there. Its a wizard of oz type thing.

[aklotz]

check out my post on page 14
first post gets the game acct gratis!

[Boldek]

I'm looking for opinions right now about Skyrim, is it as good as Halo? Do you think I could show it to a little sibling, let's say an eight year old sister?

[Insinuator]

Do you think I could show it to a little sibling, let's say an eight year old sister?

No. Definitely not.

[Boldek]

Okay. It does get kinda aggravating how every cool game that comes out always has to be M these days..

[Crendgrim]

You can't really compare Halo and Skyrim. Halo is a (very good, from what I heard) Ego-Shooter, while Skyrim is a (very good) RPG. That's a totally different thing.
I think there certainly are parts of Skyrim you could show to a 8-year-old kid; e.g. the nice landscape. I still wouldn't recommend doing so as there's a lot of blood when killing "cute" wolves...
So, if you really look for a game to show to a kid, I don't recommend going for this kind of game. Better use something like ... well, I don't know. I mainly play RPG's.

[Blarumyrran]

You can't really compare Halo and Skyrim. Halo is a (very good, from what I heard) Ego-Shooter, while Skyrim is a (very good) RPG. That's a totally different thing.

I am completely lost when people say things like this

[artisticdude]

So, if you really look for a game to show to a kid, I don't recommend going for this kind of game. Better use something like ... well, I don't know.

Like Wesnoth. Of course younger kids may have a harder time with the whole TBS genre, but my sister could play Wesnoth fairly well when she was nine, so I guess it depends on the child.

[Crendgrim]

You can't really compare Halo and Skyrim. Halo is a (very good, from what I heard) Ego-Shooter, while Skyrim is a (very good) RPG. That's a totally different thing.

I am completely lost when people say things like this

... Why? Sorry, I don't get it. Probably something obvious which I just don't see because I know what I meant to say...

[Gambit]

Okay. It does get kinda aggravating how every cool game that comes out always has to be M these days..

Because that's the gamers who have money.

[Blarumyrran]

is X as good as Y?

You can't really compare Halo and Skyrim. Halo is a (very good, from what I heard) Ego-Shooter, while Skyrim is a (very good) RPG. That's a totally different thing.

I am completely lost when people say things like this

... Why? Sorry, I don't get it. Probably something obvious which I just don't see because I know what I meant to say...

i mean - can you bring an example of a comparison of 2 real-life items that are NOT so "completely different things" that "You can't really compare" the "goodness" of them? You probably can (unless you are a very hypocritical kind of person), but then can you explain what are your general conditions for determining whether 2 real-life items are "completely different things" or not, in the context of being comparable for the question "is X as good as Y"? Or perhaps "comparableness" is a continuous, numeric value - then how do you estimate it?

Okay. It does get kinda aggravating how every cool game that comes out always has to be M these days..

Because that's the gamers who have money.

I think it's more involved than that. A large number of games are bought by parents - who are at least somewhat well-off. I'm sort of guessing at wild here, since I am not a parent myself - but I think when a parent is choosing a game to buy for his children, he chooses "a game" that fits some list of conditions: "not extremely sad", "not extremely scary", "not having a wild orgy of gore & genitalia on its cover", etc - and from those conditions emerges a rather binary threshold, and every game above that threshold is just "a game" for the parent - a commodified game; a parent probably doesn't really care about how much above that threshold a particular game is in terms of value. But for games targeted at players who buy them from their own money, no such binary threshold forms; so there is more incentive for such a game to reach for the highest quality.