Robots can't think as deeply as humans can yet, so think about what that will mean.
Advanced Squad Leader probably.
Also picking up sticks in the woods with your friends and pretending your king arthur and his round table.
That makes sense but it's not a challenge to MIT/China if they felt like it.
>AI-Proof
What does this mean? I have only ever seen AI for making art.
What do you expect to happen? AI to replace GMs or Players anytime soon? By the time AI can replace a GM it'll be smart enough that it'll hate GMing for the same reasons we do.
I expect people who aren't morons to make good posts.
>Also picking up sticks in the woods with your friends and pretending your king arthur and his round table. >That makes sense but it's not a challenge to MIT/China if they felt like it.
Elaborate on this please lol
Machines with basic autonomous machine vision and manipulation already exist. Having them larp with a behavioral model is achievable and cute, but would it be worth letting someone spend $600k-$2mln for a minor pet project?
So you are asking: >what games are not likely to be AI solvable (for what definition of solvable?) >among those games, which games are the best?
Obviously, the first question needs to be worked out because the second cannot be answered until we have a general, working definition of "solvable" agreed upon. Are we talking "plays at the level of a skilled human player" solved or are we talking "chess bot that has calculated every possible board state from here" solved?
[...]
The hardware shouldn't cost more than 10k if you're willing to power it with an extension cord.
well the technology wont be good enough to make it convincing for a long time
Well, it depends on what's meant by "think". Some machines are more learned than at least 90% of the world pop, and can provide better answers to questions, but they're hardly thinking in any real sense of thinking as cognition.
>AI-Proof
What does this mean? I have only ever seen AI for making art.
What do you expect to happen? AI to replace GMs or Players anytime soon? By the time AI can replace a GM it'll be smart enough that it'll hate GMing for the same reasons we do.
Probably any game with movement distances and flexible maneuver instead of set pathways and distances as it exponentially increases the number of moves >Do I move the squad one inch forwards, two inches, 3.5 inches, 4.25 inches, do I turn 45 degrees and drive 5 inches? Who do I shoot at? What ammunition or weapons do I use?
The issue with an AI isn’t the actual difficulty of a game it’s the mathematical complexity. To a human brain having a movement up to six inches in any direction is straightforwards to an AI it’s literally infinite options.
Until AI can develop mental flexibility like humans most non-grid based wargames are probably out of reach, this isn’t even considering list building which requires an in depth understanding not just of mechanics but of the likelihood of various situations and how units might perform in them. Lots of 40k units are good in theory but are too slow, expensive, or short ranged to be effectively utilized.
CRPGs pre-2023 aren't NLP adaptive, so they haven't replaced human DMs with IQs higher than a 5 y/o.
Bot players in RTS, action, and FPS games still do not have coverage for verbal games. A bot will get stuck in Disco Elysium, because it will run into meanings it can't process.
So you are asking: >what games are not likely to be AI solvable (for what definition of solvable?) >among those games, which games are the best?
Obviously, the first question needs to be worked out because the second cannot be answered until we have a general, working definition of "solvable" agreed upon. Are we talking "plays at the level of a skilled human player" solved or are we talking "chess bot that has calculated every possible board state from here" solved?
Machines with basic autonomous machine vision and manipulation already exist. Having them larp with a behavioral model is achievable and cute, but would it be worth letting someone spend $600k-$2mln for a minor pet project?
The hardware shouldn't cost more than 10k if you're willing to power it with an extension cord.
>Are we talking "plays at the level of a skilled human player" solved or are we talking "chess bot that has calculated every possible board state from here" solved?
Both, but start with the former, since that's what we're looking at for the foreseeable future.
interesting post for its outlines, but with severe shortcomings.
>Any game without hidden information or randomized actions can be solved, so all of those are out.
probably true
>Any game with a digital client and bot opponents has proof-of-concept of a computer playing it, which could theoretically be improved. Again, out.
this isnt as tight. just because a bot could be competitive at an early stage doesnt mean humans couldnt remain a relevant competitor. sometimes a game can reveal itself to be anthropocentric.
>As for narrative/storytelling games, chatbots exist.
im not so sure verbal games are fully solvable by ai, since im constantly running into gaps that chatbots cant solve, and theyre problems even 140 iq people would spend years struggling with. humanity might have a unique aspect that machines will never be able to replace.
It's worth looking at the history of solved games. Checkers is an interesting case, Chinook was developed by a mathematician and a pro checkers player. It was essentially a flowchart on "if x, do y". I don't think Chess or Go is solved yet per say, but there are computers that can beat most of not all human opponents.
>just because a bot could be competitive at an early stage doesnt mean humans couldnt remain a relevant competitor. sometimes a game can reveal itself to be anthropocentric.
Are there any cases of that happening? I'm interested by this concept. 15 years ago people were hyping up Go as infinitly harder for computers than Chess, yet here we are.
Verbal games are a weird case, I'll admit, especially some some of them have flexible rules. I feel like that would make it harder for a computer to competently navigate, but if the primary goal is to execute a narrative in response to a given prompt than it should be doable.
7 months ago
Anonymous
>but there are computers that can beat most of not all human opponents.
aka solved in regards to fun 🙂
7 months ago
Anonymous
>Verbal games are a weird case, I'll admit, especially some some of them have flexible rules. I feel like that would make it harder for a computer to competently navigate, but if the primary goal is to execute a narrative in response to a given prompt than it should be doable.
especially if conducted in completed languages that are walled from ai being able to get its tentacles in.
7 months ago
Anonymous
what would break these? traitors and native speakers who can train an ai in them.
but there may be languages that ai arent even able to conceptually process in.
A game with infinite starting setups can’t be solved. For example, if we gave Conway’s game of life pvp, there would be no way to solve it because you can’t quantify infinite variations.
I don't think anything is AI-proof other than games of pure chance.
The things that will probably take the longest to equal human play are either RPGs or a theoretical extremely complicated miniature wargame that likely doesn't exist.
Maybe for your berth about AI-proofness. But definitely yes for certain games reliant on mathematical randomness (AI can predict dice rolls).
>The things that will probably take the longest to equal human play are either RPGs or a theoretical extremely complicated miniature wargame that likely doesn't exist.
YES. SOMEONE ITT GETS IT, FINALLY.
"Artificial Intelligence" is a complete absurdity. It mathematically cannot exist. Even quantum computing can't emulate the discretion possible in simple biological neural systems, let alone a human brain — or any other intelligent animal. It's a matter of simple arithmetic, and you do not even remotely understand how spacially efficient a literal neural network is for what it does. None of this marketing buzzword bullshit has anything to do with this board. Go back to Ganker where they call laugh at you and stay on topic.
You could try to claim that a lab designed meat brain is artificial intelligence but at that point every live birthed human being is an artificial intelligence constructed by their mother's room from evolution's print instruction.
I don't think anything is AI-proof other than games of pure chance.
The things that will probably take the longest to equal human play are either RPGs or a theoretical extremely complicated miniature wargame that likely doesn't exist.
i know this is a shitpost, but assuming perfect play on both sides red wins on the 26th move. Like, we know exactly when the game will end. That's crazy to me.
https://connect4.gamesolver.org/
>humans and AI are on even grounds with this game and humans can always beat an AI 50% of the time
21 turns assumes perfect play on both sides. We're only on even ground if both sides play optimally, and optimal play turns the game into "first player wins."
A game with infinite starting setups can’t be solved. For example, if we gave Conway’s game of life pvp, there would be no way to solve it because you can’t quantify infinite variations.
My mistake, i should have said "any randomized elements."
No amount of research can solve the diagonalisation problem
I don't know anything about this game. What's the problem?
But wouldn't there be an ideal starting position? Unless starting location is chosen in secret, in which case the game cannot be solved (hidden information).
That being said, once starting position is established there should be a theoretical ideal move sequence.
7 months ago
Anonymous
There could be setup strategies that are more or less effective in general, but you can’t have quantifiable “perfect play”.
In a sense, it IS a hidden information game, but the information is just hidden by the singularity. You can’t compute an infinity, so even if you can solve any setup game, you cannot solve the game as a whole (because there will always be unexpected and untested setups).
7 months ago
Anonymous
Connect 4 is a game that has "solved" situations. As in, if specific conditions are met, you can consistently win the game by mechanically playing the rest of the game out a specific way. If you act accordingly starting from one of those, your opponent will not be able to get out of it.
>21 turns assumes perfect play on both sides. We're only on even ground if both sides play optimally, and optimal play turns the game into "first player wins."
That's what the 50% win rate is referencing, the human player will play 1st player 50% of games and if skilled is guaranteed to win just as the AI would be.
If player 2 plays sub optimally in any way then that just further ensures player 1's victory.
But following that logic any solved game is AI-proof, as a human player can play a game and make no errors or sub-optimal moves. Theoretically solvable games then would have the opposite situation, while top-level AI may be better than humans at Go right now, there exists a theoretical optimal game that can be discovered and executed by human players.
7 months ago
Anonymous
>while top-level AI may be better than humans at Go right now, there exists a theoretical optimal game that can be discovered and executed by human players
Agreed. Can you name any that are AI-proof for the foreseeable future?
7 months ago
Anonymous
>while top-level AI may be better than humans at Go right now, there exists a theoretical optimal game that can be discovered and executed by human players
Agreed. Can you name any that are AI-proof for the foreseeable future?
>But following that logic any solved game is AI-proof, as a human player can play a game and make no errors or sub-optimal moves. Theoretically solvable games then would have the opposite situation, while top-level AI may be better than humans at Go right now, there exists a theoretical optimal game that can be discovered and executed by human players.
There are very few other games that meet this criteria. Chess certainly does not. We're talking about games that are solvable for a human.
Connect 4 is AI proof, end of discussion.
7 months ago
Anonymous
>Connect 4 is AI proof, end of discussion.
But it's not a best game by any means, unless you count games for morons a best game.
7 months ago
Anonymous
Clearly you're not a person who enjoys fun
7 months ago
Anonymous
If you think fun means being a moron I'm not sure how you managed the rest of your posts, guy
7 months ago
Anonymous
If I need to spell it out for ya, Connect 4 is a classic fun game to play with people on occasion like tic tac toe. It absolutely one of the best games for this.
It doesn't get any more complicated than that.
7 months ago
Anonymous
But my brother cheats with AI when we play online. I don't care about drawing, but it's not building his skill level. So I need AI-proof games that are the best at developing skills that AI can't neutralize, as it did chess and go.
7 months ago
Anonymous
Brass
7 months ago
Anonymous
>Chess certainly does not.
But that's wrong you fricking moron. Any game that >has no randomized elements >has no hidden information
can be solved. Not necessarily soon, but optimal play does exist and can be discovered. Reversi, Shogi, and Hive can also be solved. Something like Poker and Mah Jong, on the other hand, cannot be solved as you don't know what pieces your opponent has and your own pieces are random. Computers can be good at them, but perfect decisions cannot be made without cheating.
Probably any game with movement distances and flexible maneuver instead of set pathways and distances as it exponentially increases the number of moves >Do I move the squad one inch forwards, two inches, 3.5 inches, 4.25 inches, do I turn 45 degrees and drive 5 inches? Who do I shoot at? What ammunition or weapons do I use?
The issue with an AI isn’t the actual difficulty of a game it’s the mathematical complexity. To a human brain having a movement up to six inches in any direction is straightforwards to an AI it’s literally infinite options.
Until AI can develop mental flexibility like humans most non-grid based wargames are probably out of reach, this isn’t even considering list building which requires an in depth understanding not just of mechanics but of the likelihood of various situations and how units might perform in them. Lots of 40k units are good in theory but are too slow, expensive, or short ranged to be effectively utilized.
Freeform movement rules and customization options are interesting points. A computer could probably play 40k but it may default to simpler movement, 16 directions would probably cover enough options for something resembling good play. List building might at first default to known packages and strong individual units, this could be easier in something like Magic.
But my brother cheats with AI when we play online. I don't care about drawing, but it's not building his skill level. So I need AI-proof games that are the best at developing skills that AI can't neutralize, as it did chess and go.
If he cheats just stop playing with him. The point of a game is to have fun.
7 months ago
Anonymous
>can be solved. Not necessarily soon, but optimal play does exist and can be discovered. Reversi, Shogi, and Hive can also be solved. Something like Poker and Mah Jong, on the other hand, cannot be solved as you don't know what pieces your opponent has and your own pieces are random. Computers can be good at them, but perfect decisions cannot be made without cheating.
You completely missed the point, I was talking about solved for humans
For example: Chess could be solved. but only an AI theoretically would be able to execute that and a human theoretically could not. However with Connect 4, it's solved and both AIs and Humans are and will always be able to execute this.
7 months ago
Anonymous
I don't think you understand what a solved game is, how much time and effort it takes to solve a game, or how shitty it is to play a game that you or the other person wins 100% of the time.
7 months ago
Anonymous
>I don't think you understand what a solved game is, how much time and effort it takes to solve a game
Yes I do:
https://en.wikipedia.org/wiki/Solved_game
>or how shitty it is to play a game that you or the other person wins 100% of the time
With Connect 4 both people will be able to win 100% of the time depending on who goes first. Connect 4 is a cute fun game on occasion and is the best game for this. It's basically one step up from Tic Tac Toe.
You have absolutely no idea what you're talking about on any level.
7 months ago
Anonymous
>With Connect 4 both people will be able to win 100% of the time depending on who goes first.
homie just flip a coin. How is playing a game with a known result any kind of fun?
7 months ago
Anonymous
You obviously have never had a good time with Connect 4
7 months ago
Anonymous
>If he cheats just stop playing with him. The point of a game is to have fun.
He cheats in chess too. I started playing word games on various apps too but he figured out how to cheat at those too. Now I make him give me book reports on Goosebumps, which he hasn't figured out a clean way to cheat at, since it necessarily requires actually confirming the contents of the book. But I'd just like to unwind with him sometimes. I'm far from a over-competitive dominator or a poor sport, so something I can equalize my over-gamifying mind so I can play with him with only a slight edge that I'll dilute occasionally..
7 months ago
Anonymous
Okay what's your response to this suggestion:
Brass
7 months ago
Anonymous
Please excuse me, I haven't had the energy to research it yet. I promise I will try to respond to your kind suggestion when I wake up.
OP has insanely specific criteria that he fully failed to mention in the opening post
He's basically in a grudge match with his cheating scumbag brother as he apparently cheats with AI at every game they play. I have no idea how that would work live, why doesn't he just play his brother live at chess? Nobody knows.
I know. Because OP is a lying homosexual and keeps switching it up.
>yeah like chess where people can solve it >wait but no not like chess because only ai can really do good at the game
He's either legitimately moronic or trolling. That much is clear. He simultaneously wants an "AI proof game" that isn't solved, but then it's okay if it's solved as long as it's just for AI, but then it can't be like that and so in.
So OP is a nogames homosexual just making shit up by ever slightly shifting the criteria back and forth. He's effort posting, but only because by adding more believability to his bait he gets far more returned in replies. I don't know how you newbies can't catch on when he's backpedaled several times on what he wants and the situation increasingly becomes more and more ridiculous.
>He's basically in a grudge match with his cheating scumbag brother as he apparently cheats with AI at every game they play.
He's 9? I work in a different state.
>OP has insanely specific criteria that he fully failed to mention in the opening post
Specificity will arrive as appropriate in the thread.
I know. Because OP is a lying homosexual and keeps switching it up.
>yeah like chess where people can solve it >wait but no not like chess because only ai can really do good at the game
He's either legitimately moronic or trolling. That much is clear. He simultaneously wants an "AI proof game" that isn't solved, but then it's okay if it's solved as long as it's just for AI, but then it can't be like that and so in.
So OP is a nogames homosexual just making shit up by ever slightly shifting the criteria back and forth. He's effort posting, but only because by adding more believability to his bait he gets far more returned in replies. I don't know how you newbies can't catch on when he's backpedaled several times on what he wants and the situation increasingly becomes more and more ridiculous.
>I know. Because OP is a lying homosexual and keeps switching it up.
I'm not every poster, guy. You're weird.
>>He's basically in a grudge match
It's far from a grudge match. He's young, and I think it's funny. He's a pretty smart kid who's skipped three grades, but he's just misled in his basic approach to games.
Alright the game anon mentioned is provisionally AI-proof, but that appears to be more a matter of circumstance. I just bought copies for our accounts, and I'll have to test it and see how time-proof it is.
None.
Robots can't think as deeply as humans can yet, so think about what that will mean.
That makes sense but it's not a challenge to MIT/China if they felt like it.
I expect people who aren't morons to make good posts.
Buddy, either explain what you actually fear or frick off. Because the answer is all TTRPGs for the foreseeable future.
>just intuition what I wanted my thread to be about because I can't be assed to elaborate
No, frick you
>Also picking up sticks in the woods with your friends and pretending your king arthur and his round table.
>That makes sense but it's not a challenge to MIT/China if they felt like it.
Elaborate on this please lol
Machines with basic autonomous machine vision and manipulation already exist. Having them larp with a behavioral model is achievable and cute, but would it be worth letting someone spend $600k-$2mln for a minor pet project?
Now let's market this, like maybe rent six of these things for a weekend
well the technology wont be good enough to make it convincing for a long time
>Robots can't think as deeply as humans can
Robots already think deeper than a double digit percentage of the human race
that you really believe this, puts you in the percentage
Well, it depends on what's meant by "think". Some machines are more learned than at least 90% of the world pop, and can provide better answers to questions, but they're hardly thinking in any real sense of thinking as cognition.
Advanced Squad Leader probably.
Also picking up sticks in the woods with your friends and pretending your king arthur and his round table.
y u post benas
>AI-Proof
What does this mean? I have only ever seen AI for making art.
What do you expect to happen? AI to replace GMs or Players anytime soon? By the time AI can replace a GM it'll be smart enough that it'll hate GMing for the same reasons we do.
I assume he means things like Stockfish and AlphaGo
Yes hence the OP picrel that filters morons.
like
m not rtetrad
Probably any game with movement distances and flexible maneuver instead of set pathways and distances as it exponentially increases the number of moves
>Do I move the squad one inch forwards, two inches, 3.5 inches, 4.25 inches, do I turn 45 degrees and drive 5 inches? Who do I shoot at? What ammunition or weapons do I use?
The issue with an AI isn’t the actual difficulty of a game it’s the mathematical complexity. To a human brain having a movement up to six inches in any direction is straightforwards to an AI it’s literally infinite options.
Until AI can develop mental flexibility like humans most non-grid based wargames are probably out of reach, this isn’t even considering list building which requires an in depth understanding not just of mechanics but of the likelihood of various situations and how units might perform in them. Lots of 40k units are good in theory but are too slow, expensive, or short ranged to be effectively utilized.
AI WILL replace both DMs and PLAYERS one day.
for many types of games, certainly. but not all games, which humans will shift their emphasis to.
It's already happened. CRPGs replaced the DM and aimbots replaced the players
CRPGs pre-2023 aren't NLP adaptive, so they haven't replaced human DMs with IQs higher than a 5 y/o.
Bot players in RTS, action, and FPS games still do not have coverage for verbal games. A bot will get stuck in Disco Elysium, because it will run into meanings it can't process.
Players will get replaced first
>Because the answer is all TTRPGs for the foreseeable future.
1. I asked for the best. Reread the OP, moron.
2. LOOK AT THE OP PICREL.
So you are asking:
>what games are not likely to be AI solvable (for what definition of solvable?)
>among those games, which games are the best?
Obviously, the first question needs to be worked out because the second cannot be answered until we have a general, working definition of "solvable" agreed upon. Are we talking "plays at the level of a skilled human player" solved or are we talking "chess bot that has calculated every possible board state from here" solved?
The hardware shouldn't cost more than 10k if you're willing to power it with an extension cord.
>Are we talking "plays at the level of a skilled human player" solved or are we talking "chess bot that has calculated every possible board state from here" solved?
Both, but start with the former, since that's what we're looking at for the foreseeable future.
Any game without hidden information or randomized actions can be solved, so all of those are out.
Any game with a digital client and bot opponents has proof-of-concept of a computer playing it, which could theoretically be improved. Again, out.
As for narrative/storytelling games, chatbots exist.
The game you're looking for is Calvinball.
interesting post for its outlines, but with severe shortcomings.
>Any game without hidden information or randomized actions can be solved, so all of those are out.
probably true
>Any game with a digital client and bot opponents has proof-of-concept of a computer playing it, which could theoretically be improved. Again, out.
this isnt as tight. just because a bot could be competitive at an early stage doesnt mean humans couldnt remain a relevant competitor. sometimes a game can reveal itself to be anthropocentric.
>As for narrative/storytelling games, chatbots exist.
im not so sure verbal games are fully solvable by ai, since im constantly running into gaps that chatbots cant solve, and theyre problems even 140 iq people would spend years struggling with. humanity might have a unique aspect that machines will never be able to replace.
It's worth looking at the history of solved games. Checkers is an interesting case, Chinook was developed by a mathematician and a pro checkers player. It was essentially a flowchart on "if x, do y". I don't think Chess or Go is solved yet per say, but there are computers that can beat most of not all human opponents.
>just because a bot could be competitive at an early stage doesnt mean humans couldnt remain a relevant competitor. sometimes a game can reveal itself to be anthropocentric.
Are there any cases of that happening? I'm interested by this concept. 15 years ago people were hyping up Go as infinitly harder for computers than Chess, yet here we are.
Verbal games are a weird case, I'll admit, especially some some of them have flexible rules. I feel like that would make it harder for a computer to competently navigate, but if the primary goal is to execute a narrative in response to a given prompt than it should be doable.
>but there are computers that can beat most of not all human opponents.
aka solved in regards to fun 🙂
>Verbal games are a weird case, I'll admit, especially some some of them have flexible rules. I feel like that would make it harder for a computer to competently navigate, but if the primary goal is to execute a narrative in response to a given prompt than it should be doable.
especially if conducted in completed languages that are walled from ai being able to get its tentacles in.
what would break these? traitors and native speakers who can train an ai in them.
but there may be languages that ai arent even able to conceptually process in.
A game with infinite starting setups can’t be solved. For example, if we gave Conway’s game of life pvp, there would be no way to solve it because you can’t quantify infinite variations.
Yes you theoretically could. You must not be familiar with the last 20 years of research.
No amount of research can solve the diagonalisation problem
Haha, another moron I filtered using my clever bean trick.
AI isn't real, you fricking moron.
I'm OP, duh. So far.
Maybe for your berth about AI-proofness. But definitely yes for certain games reliant on mathematical randomness (AI can predict dice rolls).
>The things that will probably take the longest to equal human play are either RPGs or a theoretical extremely complicated miniature wargame that likely doesn't exist.
YES. SOMEONE ITT GETS IT, FINALLY.
"Artificial Intelligence" is a complete absurdity. It mathematically cannot exist. Even quantum computing can't emulate the discretion possible in simple biological neural systems, let alone a human brain — or any other intelligent animal. It's a matter of simple arithmetic, and you do not even remotely understand how spacially efficient a literal neural network is for what it does. None of this marketing buzzword bullshit has anything to do with this board. Go back to Ganker where they call laugh at you and stay on topic.
You could try to claim that a lab designed meat brain is artificial intelligence but at that point every live birthed human being is an artificial intelligence constructed by their mother's room from evolution's print instruction.
I don't think anything is AI-proof other than games of pure chance.
The things that will probably take the longest to equal human play are either RPGs or a theoretical extremely complicated miniature wargame that likely doesn't exist.
This is the only game I care to mention that is AI proof for all time and eternity
Seethe robots!
i know this is a shitpost, but assuming perfect play on both sides red wins on the 26th move. Like, we know exactly when the game will end. That's crazy to me.
https://connect4.gamesolver.org/
21 moves. My bad, I haven't thought about this in a few years
That's what i mean by it being AI proof, humans and AI are on even grounds with this game and humans can always beat an AI 50% of the time
>humans and AI are on even grounds with this game and humans can always beat an AI 50% of the time
21 turns assumes perfect play on both sides. We're only on even ground if both sides play optimally, and optimal play turns the game into "first player wins."
My mistake, i should have said "any randomized elements."
I don't know anything about this game. What's the problem?
You don’t need randomness to have infinite starting conditions.
But wouldn't there be an ideal starting position? Unless starting location is chosen in secret, in which case the game cannot be solved (hidden information).
That being said, once starting position is established there should be a theoretical ideal move sequence.
There could be setup strategies that are more or less effective in general, but you can’t have quantifiable “perfect play”.
In a sense, it IS a hidden information game, but the information is just hidden by the singularity. You can’t compute an infinity, so even if you can solve any setup game, you cannot solve the game as a whole (because there will always be unexpected and untested setups).
Connect 4 is a game that has "solved" situations. As in, if specific conditions are met, you can consistently win the game by mechanically playing the rest of the game out a specific way. If you act accordingly starting from one of those, your opponent will not be able to get out of it.
>21 turns assumes perfect play on both sides. We're only on even ground if both sides play optimally, and optimal play turns the game into "first player wins."
That's what the 50% win rate is referencing, the human player will play 1st player 50% of games and if skilled is guaranteed to win just as the AI would be.
If player 2 plays sub optimally in any way then that just further ensures player 1's victory.
But following that logic any solved game is AI-proof, as a human player can play a game and make no errors or sub-optimal moves. Theoretically solvable games then would have the opposite situation, while top-level AI may be better than humans at Go right now, there exists a theoretical optimal game that can be discovered and executed by human players.
>while top-level AI may be better than humans at Go right now, there exists a theoretical optimal game that can be discovered and executed by human players
Agreed. Can you name any that are AI-proof for the foreseeable future?
>But following that logic any solved game is AI-proof, as a human player can play a game and make no errors or sub-optimal moves. Theoretically solvable games then would have the opposite situation, while top-level AI may be better than humans at Go right now, there exists a theoretical optimal game that can be discovered and executed by human players.
There are very few other games that meet this criteria. Chess certainly does not. We're talking about games that are solvable for a human.
Connect 4 is AI proof, end of discussion.
>Connect 4 is AI proof, end of discussion.
But it's not a best game by any means, unless you count games for morons a best game.
Clearly you're not a person who enjoys fun
If you think fun means being a moron I'm not sure how you managed the rest of your posts, guy
If I need to spell it out for ya, Connect 4 is a classic fun game to play with people on occasion like tic tac toe. It absolutely one of the best games for this.
It doesn't get any more complicated than that.
But my brother cheats with AI when we play online. I don't care about drawing, but it's not building his skill level. So I need AI-proof games that are the best at developing skills that AI can't neutralize, as it did chess and go.
Brass
>Chess certainly does not.
But that's wrong you fricking moron. Any game that
>has no randomized elements
>has no hidden information
can be solved. Not necessarily soon, but optimal play does exist and can be discovered. Reversi, Shogi, and Hive can also be solved. Something like Poker and Mah Jong, on the other hand, cannot be solved as you don't know what pieces your opponent has and your own pieces are random. Computers can be good at them, but perfect decisions cannot be made without cheating.
Freeform movement rules and customization options are interesting points. A computer could probably play 40k but it may default to simpler movement, 16 directions would probably cover enough options for something resembling good play. List building might at first default to known packages and strong individual units, this could be easier in something like Magic.
If he cheats just stop playing with him. The point of a game is to have fun.
>can be solved. Not necessarily soon, but optimal play does exist and can be discovered. Reversi, Shogi, and Hive can also be solved. Something like Poker and Mah Jong, on the other hand, cannot be solved as you don't know what pieces your opponent has and your own pieces are random. Computers can be good at them, but perfect decisions cannot be made without cheating.
You completely missed the point, I was talking about solved for humans
For example: Chess could be solved. but only an AI theoretically would be able to execute that and a human theoretically could not. However with Connect 4, it's solved and both AIs and Humans are and will always be able to execute this.
I don't think you understand what a solved game is, how much time and effort it takes to solve a game, or how shitty it is to play a game that you or the other person wins 100% of the time.
>I don't think you understand what a solved game is, how much time and effort it takes to solve a game
Yes I do:
https://en.wikipedia.org/wiki/Solved_game
>or how shitty it is to play a game that you or the other person wins 100% of the time
With Connect 4 both people will be able to win 100% of the time depending on who goes first. Connect 4 is a cute fun game on occasion and is the best game for this. It's basically one step up from Tic Tac Toe.
You have absolutely no idea what you're talking about on any level.
>With Connect 4 both people will be able to win 100% of the time depending on who goes first.
homie just flip a coin. How is playing a game with a known result any kind of fun?
You obviously have never had a good time with Connect 4
>If he cheats just stop playing with him. The point of a game is to have fun.
He cheats in chess too. I started playing word games on various apps too but he figured out how to cheat at those too. Now I make him give me book reports on Goosebumps, which he hasn't figured out a clean way to cheat at, since it necessarily requires actually confirming the contents of the book. But I'd just like to unwind with him sometimes. I'm far from a over-competitive dominator or a poor sport, so something I can equalize my over-gamifying mind so I can play with him with only a slight edge that I'll dilute occasionally..
Okay what's your response to this suggestion:
Please excuse me, I haven't had the energy to research it yet. I promise I will try to respond to your kind suggestion when I wake up.
What are the best RPG and wargames for the OP question?
52 pickup
AI solved. They used specialized rubber manipulators.
AI only made e.g. chess better. No reason to try to avoid it.
OP has insanely specific criteria that he fully failed to mention in the opening post
He's basically in a grudge match with his cheating scumbag brother as he apparently cheats with AI at every game they play. I have no idea how that would work live, why doesn't he just play his brother live at chess? Nobody knows.
I know. Because OP is a lying homosexual and keeps switching it up.
>yeah like chess where people can solve it
>wait but no not like chess because only ai can really do good at the game
He's either legitimately moronic or trolling. That much is clear. He simultaneously wants an "AI proof game" that isn't solved, but then it's okay if it's solved as long as it's just for AI, but then it can't be like that and so in.
So OP is a nogames homosexual just making shit up by ever slightly shifting the criteria back and forth. He's effort posting, but only because by adding more believability to his bait he gets far more returned in replies. I don't know how you newbies can't catch on when he's backpedaled several times on what he wants and the situation increasingly becomes more and more ridiculous.
>He's basically in a grudge match with his cheating scumbag brother as he apparently cheats with AI at every game they play.
He's 9? I work in a different state.
>OP has insanely specific criteria that he fully failed to mention in the opening post
Specificity will arrive as appropriate in the thread.
>I know. Because OP is a lying homosexual and keeps switching it up.
I'm not every poster, guy. You're weird.
>>He's basically in a grudge match
It's far from a grudge match. He's young, and I think it's funny. He's a pretty smart kid who's skipped three grades, but he's just misled in his basic approach to games.
Look into gambling.
Alright the game anon mentioned is provisionally AI-proof, but that appears to be more a matter of circumstance. I just bought copies for our accounts, and I'll have to test it and see how time-proof it is.