Is Artificial Intelligence Superior to Human Intelligence?
- This topic has 21 replies, 1 voice, and was last updated 1 year, 11 months ago by
.
-
Posts
-
Is it feasible to build an AI for Age of Empires II that can outperform human players without extra resources or an unfair advantage?
Although this has already been accomplished in chess, Age of Empires II is much more complex and dynamic.
AI is better at resource management than most humans IMO.
But they’re terrible at making macro decisions such as where to attack and when, taking advantage of tech timings and choosing fights, and choosing when to add TCs vs push army.
I don’t think the AI is ever going to be improved to the point that it will feel the same as a human player.
It would absolutely dominate ranged battles and do some ungodly things with monks.
Not sure about it’s ability to predict movements to attack ground with mangos, but aside from that, absolutely unstoppable
AI can really kick human players asses in CSGO =(
Du..
It would be a challenge with a cap on APM
Yes
They did this for Dota.
It was AI that played milions of matches and learned from them.
it even made its own meta game.
Pro had rly hard time to beat it and it took a lot matches to beat the AI actually.
Rly scary how effecient they were.
If i remember correctly pros even took some little things from it.
It exists in Starcraft 2 and I reckon it’s easier to do it in aoe2 than in sc2
I reckon that an AI winning by out strategizing a human its nearly impossible.
Human can always jebait the bot, like showing a couple of archers to force a reaction and then counter playing with scouts, for example.
Winning by out microing, may be “simplier” to code, but heavy resource consuming, making it hard to run successfully in mid spec computers.
Edit: this is a very interesting topic.
Would be awesome to hear from AI developers!
I’d say that’s pretty darn close.
I would imagine that AI would beat at least 90% of humans lol and the rest would struggle against it.
Idk if it has been openly tested against humans tho.
Absolutely.
In a game that can be decided in a few big fights, even something like an improved Rehoboam AI will wipe the floor with pro players.
While other commenters have talked about the microing superiority of a bot that does not have human stress responses or input delay, the macro aspect of an efficient AI can’t be overlooked either.
They will always nail build orders and with enough machine learning, be able to respond effectively to deviations as well.
Imagine teaching an AI how to wall trap every time, or use demo ships for maximum impact.
We (especially pro players) rely on heuristics ourselves, and those patterns can be conceivably taught to an AI too.
Many complex games, most notably starcraft have created AI’s that outperform top humans, even with apm and reaction time limits.
It is certainly possible to create a similar AI for aoe2, but it would take the same amount of work that it did for these other games, namely a team of professionals workings for months or years.
The extreme AI is better than half the ranked playerbase, so yes!
As others have mentioned, AIs have been made that crush humans in other esports like Starcraft 2 and Dota 2, and there is no reason to think that AoE2 would be any different.
But, in my humble opinion, these AI were not competing under fair conditions.
What do I mean by this?
Well, let us think about what our brain has to do in order to do something in a game.
First, we get some stimulus, either something on the screen or an audio clue.
The brain then has to decode what this stimulus is (this can go wrong all the time, we may miss something on screen, we may misunderstand what something is etc), the brain then has to decide what to do in the game (this is the only part that the AI does).
Then the brain has to figure out what nerves to operate such as to make our hands do the actions to execute the plan (which is something that goes wrong all the time in the form of misclicks, mis-hotkeying, being slow, etc).
That we don’t do two or indeed all three of these steps consciously doesn’t matter, after all, *nothing* the AI does is conscious.
If the game was to be fair between the AI and the Humans, the AI should get all its stimulus from video and audio like us humans, and control a physical rig of some sort that would then operate a mouse and keyboard like us humans.
Only then would I say that the AI actually beat the humans at the game.
This does however open up a further quandary, of how good should the AI’s new hardware be.
Is it ok if the AI’s camera is superior to the human eye?
Perhaps the physical rig is purpose built to operate a mouse and keyboard and so better at it than human hands are?
I admit that I have no answer to this, but I stand by my principal that the AIs in question were doing but a small fraction of the actions our brains have to do to play.
Now, from the point of view of computer science this might be beside the point, and that’s perfectly fair.
In truth, I’m not sure what the point of making AI play games is besides being an eloborate marketing campaign, showing that a company has the ability to create great AI.
As a disclaimer, I’m in no way claiming that an AI wouldn’t be able to do any of these steps.
In fact, I don’t think there is any ability or skill that humans have that AI won’t eventually be able to do much better than we are, and I think given sufficient time humans will become obsolete, incapable of contributing to society except in the form of getting out of the way of much more capable machines.
- You must be logged in to reply to this topic.
