I challenge someone to develop AI that able to win 5 vs 5 game (MOBA legend and DOTA) in international tournament.
LOL? you just came out from cave bro? There's already OpenAI that defeats the 2 times dota world champion
Also every now and then, there'll always be stupid people like TS that said AI can't do much, same goes to the guy that said AI cannot defeat Garry Kasparov.
AI already helped robots like Boston Dynamics run and do backflip, next they'll be charging in your doors with M17.
Game developer can actually produce unbeatable A.I at all cost if they want to... Then no one will be playing the game because it is unbeatable, is not good for their business.
Developer can give AI map hack, it know where player are and running away when player engage it and hunt player when player hunting in forest / hunting roshan and equip the AI with 100% last kill rate and 100 % denied rate, it can calculate accurately the damage and creep hit point. That is practically a cheat for the computer. if the developer want to do that.
maphack for AI isn't the best argument here, I suggest you understand what AI truly does in youtube, search for Two Minute Papers.
Machine learning simply "learns" from preexisting data and does some statistical analysis and tells you the best answer based on historical data.
Can a ML model generate its own data and learn from it? Can a ML model provide you with creative solutions that is not based on old data? Put it this way. Create two chess ML models and let them play against each other and see how far they can go. After awhile, let the model play against a human chess player. I do not think they would even remotely come close to beating a good human chess player.
Machine learning does not have "common-sense". It's just statistics.
ever heard of unsupervised and reinforcement learning? it doesn't necessarily requires a subset of data and mimic it.
Can't blame him. Malaysia exposure to this kind of high tech field is really terrible. The last paper i read, 1 lab already getting some impressive results.
I know some dude in US/Canada highschool can complete our UM's Master Degree thesis, what more can I say about this country, if you are active in oversea forums, a lot of scholars think our country's Master cert is toilet paper.
Right, but how about economic? With 100% last kill and denied rate, it can earn money then place many ward, and with that, it's not consider as map hack, but AI can see the map and it can pick its fight, it will decide when to run away and when it want to fight. Sabotaging player in forest etc.
Finding the pathing and timing will be essential. No human can calculate as accurately as an AI. It can also calculate the cooldown of each spells and damage it can do to human character and totaled up the spell combo damage.
I think developer already can develop this, but really, no one will want to play against such bot.
what are you talking about dude, there's already an AI called OpenAI that defeated the world champion, learn to google la, dont be close minded and expect people to spoonfeed
I actually watched the match live and IIRC the hero pool was very limited. Like 15 instead of 100+. Also they essentially have "macros". If you play Meepo you would know what I mean. Human players aren't allowed to use macro keys too (competitively speaking). I also remember OpenAI initially are given "semi-maphack" and humans of course are limited to what they can see on the screen at any given instant. They can't see the entire visible map.
Great achievement nonetheless but I wouldn't call it "beating human players" yet for dota 2
So you think the improvement has stopped? In the next ten or hundred of years, in the field of AI, we’ll still be where we are today? There’s no more OpenAI for Dota anymore since two years ago because the developers have proven their point, not because they couldn’t expand the AI anymore for Dota. Look at Alpha GO Zero, the possibilities are endless, modern computing has only been around for 10-20 years, imagine what it’ll do in the next ten years, the growth and research will only go exponential, companies are looking ways to optimise, hence AI and automation, the Industry 4.0, its real whether you believe it or not.
Computers cannot think by themselves so true AI doesn't exist. Any ML model out there still requires human to help them, one way or another.
Can a ML model write a good and original poetry that makes sense? When it writes it, does it understand what it's writing? Does it have any knowledge of the words its putting out? Or are they just making up sentences based on probabilities? Sure I've seen some poetry generated by some ML models out there, but does it know and understand what it's writing?
This is what I meant when I say true AI doesn't exist. It lacks the sense and touch of reality.
There are models for creating songs, create articles, etc, and they’re highly accurate, some of the articles you read online are generated by AI, you just didn’t know it. I highly suggest you watch couple of videos from Two Minutes Paper at YouTube, that’ll change your mind.