| Forum Index > SC2 General |
|
Elentos
55555 Posts
On March 28 2016 19:20 lichter wrote: if he actually watches too many replays and learns to type more than glhf and gg then he definitely will get DQ'd by kespa If nothing else, this would be hilarious. | ||
|
Mendelfist
Sweden356 Posts
On March 28 2016 19:19 Elentos wrote: All this would lead to is a squad of KeSPA referees running in and disqualifying the AI. It would foresee this move of course, so it would start by BM-ing the referees out of the building. | ||
|
sharkie
Austria18511 Posts
On March 28 2016 19:32 Mendelfist wrote: As someone who has played chess (no rank), Go (10 kyu) and Starcraft (platinum) at low amateur level I say you are wrong. At the strategic level Go is enormously more complex and deep than both chess and Starcraft. There is absolutely no comparison. Go completely dwarfs any game that I have come in contact with. Measuring complexity in "number of permutations" is not meaningful. How many permutations are there in a game of billiard? Do you think it would be hard for a computer? you should ignore people like them, they actually believe starcraft is the most complex game ever | ||
|
dust7
199 Posts
On March 28 2016 14:22 ThunderBum wrote: I would like to see the ai learn without any APM restraints to see what kind of micro it learns is the best. Will it learn when it's best to stand and fight for a trade and when it's best to run away? I think limiting the APM puts a limitation on the ai that it shouldn't have. The question shouldn't be whether or not a human can beat a handicapped ai, it should be wether ai can leverage all of it's advantages to beat humans just as how humans will leverage their advantages to beat the ai. But it is obvious that no human could beat an AI with even subpar decisionmaking and some logic to handle abusive corner-cases if it had perfect micro, macro and map awareness, microing every single marine at ten places at once. That's not interesting. It is about as boring as installing a wall behind a tennis net and let it compete against a human player. We already know who wins, and why. The interesting part is the question whether humans can construct an AI that OUTPLAYS a human player due to better decisionmaking when it has roughly the same limitations as a human player (i.e. not being able to see every screen at once, apm cap, etc.). It is not about being fair or unfair, it is about creating interesting scenarios that actually test the wit of the AI. Testing whether a computer can have inhuman micro and macro is pointless, we know the answer. | ||
|
Dumbledore
Sweden725 Posts
On March 28 2016 17:09 spinesheath wrote: China already has a supercomputer that has about the same computing power as the estimated computing power of the human brain. In about 10-20 years this computing power will cost like a couple thousand dollars (this is just based on a continuation of moore's law, which has held true for a very long time). You can easily code an AI to take say 1 second to come up with a bunch of potential moves and then pick the one that looks the most promising so far. By no means does it have to evaluate all possiblities or any such nonsense. Obviously you can tinker with that decision making process any way you want; there could be thousands of decisions for different aspects of the game being evaluated in parallel. Also humans don't think quickly when compared to computers. Brain cells operate at a much lower frequency as a modern cpu and signal transmission between cells is much slower than on a cpu. Moore's law is not holding true anymore. | ||
|
_fool
Netherlands678 Posts
On March 28 2016 19:32 Mendelfist wrote: As someone who has played chess (no rank), Go (10 kyu) and Starcraft (platinum) at low amateur level I say you are wrong. At the strategic level Go is enormously more complex and deep than both chess and Starcraft. There is absolutely no comparison. Go completely dwarfs any game that I have come in contact with. Measuring complexity in "number of permutations" is not meaningful. How many permutations are there in a game of billiard? Do you think it would be hard for a computer? Well put! Wanted to respond myself, but you said it better than I could've  | ||
|
Mikau
Netherlands1446 Posts
On March 28 2016 14:35 lpunatic wrote: BW AIs with no APM limits still can't beat decent humans, despite super muta micro and whatnot. That's the inbuilt BW AI, which uses a tiny fraction of the processing power AlphaGo/Star uses. You can't even remotely compare the two. | ||
|
LetaBot
Netherlands557 Posts
| ||
|
spinesheath
Germany8679 Posts
It's still close enough to true; computing power still grows exponentially. And in any case companies like Google will have easy access to human brain level computing power within 10-20 years. On March 28 2016 20:07 Mikau wrote: That's the inbuilt BW AI, which uses a tiny fraction of the processing power AlphaGo/Star uses. You can't even remotely compare the two. He's talking about third party AIs. Those are much more elaborate than the simple built in AIs. | ||
|
royalroadweed
United States8301 Posts
| ||
|
Mikau
Netherlands1446 Posts
On March 28 2016 20:16 spinesheath wrote: It's still close enough to true; computing power still grows exponentially. And in any case companies like Google will have easy access to human brain level computing power within 10-20 years. He's talking about third party AIs. Those are much more elaborate than the simple built in AIs. Yet they still don't run on20 CPUs like AlphaGo did. You really can't compare them. | ||
Poopi
France12904 Posts
| ||
|
InfCereal
Canada1759 Posts
Wasn't it taking alphago minutes to decide moves in go? If Starcraft is even a fraction as complex as go, wouldn't its apm actually be quite slow because it's trying to figure out how to win every game? | ||
|
dust7
199 Posts
On March 28 2016 21:24 InfCereal wrote: Would the AI even have unbelievable APM? Wasn't it taking alphago minutes to decide moves in go? If Starcraft is even a fraction as complex as go, wouldn't its apm actually be quite slow because it's trying to figure out how to win every game? I think the point is that there's not much decision making going on when executing close to perfect micro - think Automaton 2000 Marines vs Banelings or Tank pickup micro. This micro can be pretty much hardcoded and the AI would be instructed to execute it whenever possible if it had unlimited APM. Perfect micro for any given situation due to making actual decisions would be computationally expensive, but inhuman, extremely good general micro algorithms would be good enough. | ||
|
DuckloadBlackra
225 Posts
| ||
|
kingjames01
Canada1603 Posts
On March 28 2016 17:49 DinosaurPoop wrote: Why do we need an apm limit you guys seriously overestimate the power of perfect micro. Have you ever watched a video of a StarCraft II Micro bot? I think you are seriously underestimating the power of perfect micro. | ||
|
kingjames01
Canada1603 Posts
On March 28 2016 19:15 MockHamill wrote: If the cap the APM to that of a pro player I do not think DeepMind has a chance. Starcraft is so much more complex than Go it is absurd. Basically the amount of permutations in the first 3 minutes of Starcraft is larger then any match of Go. What are you even saying here? How many permutations do you think there are in the first 3 minutes of StarCraft? Are you including the co-ordinates of each of your units in this count? Be realistic. Sort board states into discrete groups and do another count. Do you have ANY idea how many possible board states there are in a single match of Go? Possible board states in Chess: 10^120 Possible board states in Go: 10^761 | ||
|
kingjames01
Canada1603 Posts
On March 28 2016 20:45 Mikau wrote: Yet they still don't run on20 CPUs like AlphaGo did. You really can't compare them. AlphaGo used 1920 CPUs and 280 GPUs against Lee Sedol. | ||
|
DinoMight
United States3725 Posts
When the machine played Jeopardy they made it think and gave it a human reaction time to hitting the buzzer, otherwise it would have just destroyed everyone by buzzing immediately on every question and then thinking of the answer in the next second or two. | ||
|
SharpFlex
Germany23 Posts
you heard it here first | ||
| ||
|
|
Wardi Open
Wardi Open
Replay Cast
WardiTV Korean Royale
Replay Cast
Kung Fu Cup
Classic vs Solar
herO vs Cure
Reynor vs GuMiho
ByuN vs ShoWTimE
Tenacious Turtle Tussle
The PondCast
RSL Revival
Solar vs Zoun
MaxPax vs Bunny
Kung Fu Cup
[ Show More ] |
|
|
EPT
