In 2016 the most advance AI, using deep learning and reinforcement learning, called AlphaGo defeated the best human Go player. If Google's AI can defeat human's at the game of Go, then why can't it play complex video games like call of duty, Halo or Zelda?
1. Google's Deep Q network (reinforcement learning using a deep neural network)
2. IBM Watson
3. Apple's Siri.
None of the current AI in 2016 can play the Call of duty. That includes the 3 most popular Artificial Intelligence above. The question is why? It's because the current AI:
1. can't do addition and subtraction or basic math.
2. can't understand natural language.
3. can't do recursive tasks (from simple to very complex)
4. can't make intelligent decisions.
5. can't generate common sense knowledge.
6. can't play an unknown video game without prior training or have the ability to play similar games.
7. can't navigate in an unknown maze.
8. can't learn, identify, or recall goals and rules of a game or self-discover rules and goals of an unknown game.
The list goes on and on. To illustrate my point i will give one example from the call of duty. Imagine the robot is playing the call of duty and he is in a situation where the player has 2 bullets; and 5 enemy soldiers are surrounding the player (the robot). The robot has to add and subtract because he has to know that 5 minus 2 is 3 and that means the robot needs 3 more bullets in order to take out 5 enemy soldiers.
Next, the robot has to understand common sense knowledge. 1 bullet isn't enough to kill 1 enemy soldier. Sometimes the player might miss a shot or it takes 3-4 bullets to kill one enemy. This is where the robot has to understand common sense knowledge and know that 5 bullets isn't enough to kill 5 enemy soldiers. He needs approximately 5 X 5 = 25 bullets in order to kill 5 enemy soldiers.
The 3 AI programs mentioned above doesn't have these 2 basic skills (do simple math and generate common sense knowledge). Their respective Artificial Intelligence can't add, subtract, or understand simple concepts like gravity.
Next, in order for the robot to play the call of duty, it has to understand natural language (understand English sentences and paragraphs). In the game of Call of duty, at the beginning of each level, a mission is given to the player via English sentences. The robot has to read and understand the sentences. The sentences might tell the player to rescue a hostage from building5 or protect hostages from enemy gunfire, etc. The robot has to understand natural language in order to play the game of call of duty, which is something Google, IBM and Apple's AI can't do.
Next, we have complex decision making. If the player (the robot) is on a building and he had to make a decision to jump off the building or engage enemy soldiers, it will take someone very intelligent to survive that kind of situation. The robot certainly can't jump off the building because gravity will kill the robot (common sense knowledge). Engaging the enemy is also suicide because the robot is outnumbered. A smart robot will try to hide or jump to the second level of the building in order to survive. This type of decision making is very complex and requires many basic skills (like math, knowledge about gravity, planning, and logical inference).
There you have it folks. The reason all the Artificial Intelligence in 2016 can't play games like doom or call of duty is because they can't add, subtract, understand grade school English, lack of basic decision making skills, or understand simple concepts like gravity. Humanity is a long way from building AI that is comparable to human intelligence (at a college level).
To further illustrate my point I will show a video on how my Artificial intelligence plays Halo. Watch the video and see what knowledge and thinking is required to play a complex game like Halo. Numerous patents have been filed on this invention starting from 2006, called Human Level AI (aka artificial general intelligence).
all comments are welcomed
http://www.humanlevelartificialintelligence.com
1. Google's Deep Q network (reinforcement learning using a deep neural network)
2. IBM Watson
3. Apple's Siri.
None of the current AI in 2016 can play the Call of duty. That includes the 3 most popular Artificial Intelligence above. The question is why? It's because the current AI:
1. can't do addition and subtraction or basic math.
2. can't understand natural language.
3. can't do recursive tasks (from simple to very complex)
4. can't make intelligent decisions.
5. can't generate common sense knowledge.
6. can't play an unknown video game without prior training or have the ability to play similar games.
7. can't navigate in an unknown maze.
8. can't learn, identify, or recall goals and rules of a game or self-discover rules and goals of an unknown game.
The list goes on and on. To illustrate my point i will give one example from the call of duty. Imagine the robot is playing the call of duty and he is in a situation where the player has 2 bullets; and 5 enemy soldiers are surrounding the player (the robot). The robot has to add and subtract because he has to know that 5 minus 2 is 3 and that means the robot needs 3 more bullets in order to take out 5 enemy soldiers.
Next, the robot has to understand common sense knowledge. 1 bullet isn't enough to kill 1 enemy soldier. Sometimes the player might miss a shot or it takes 3-4 bullets to kill one enemy. This is where the robot has to understand common sense knowledge and know that 5 bullets isn't enough to kill 5 enemy soldiers. He needs approximately 5 X 5 = 25 bullets in order to kill 5 enemy soldiers.
The 3 AI programs mentioned above doesn't have these 2 basic skills (do simple math and generate common sense knowledge). Their respective Artificial Intelligence can't add, subtract, or understand simple concepts like gravity.
Next, in order for the robot to play the call of duty, it has to understand natural language (understand English sentences and paragraphs). In the game of Call of duty, at the beginning of each level, a mission is given to the player via English sentences. The robot has to read and understand the sentences. The sentences might tell the player to rescue a hostage from building5 or protect hostages from enemy gunfire, etc. The robot has to understand natural language in order to play the game of call of duty, which is something Google, IBM and Apple's AI can't do.
Next, we have complex decision making. If the player (the robot) is on a building and he had to make a decision to jump off the building or engage enemy soldiers, it will take someone very intelligent to survive that kind of situation. The robot certainly can't jump off the building because gravity will kill the robot (common sense knowledge). Engaging the enemy is also suicide because the robot is outnumbered. A smart robot will try to hide or jump to the second level of the building in order to survive. This type of decision making is very complex and requires many basic skills (like math, knowledge about gravity, planning, and logical inference).
There you have it folks. The reason all the Artificial Intelligence in 2016 can't play games like doom or call of duty is because they can't add, subtract, understand grade school English, lack of basic decision making skills, or understand simple concepts like gravity. Humanity is a long way from building AI that is comparable to human intelligence (at a college level).
To further illustrate my point I will show a video on how my Artificial intelligence plays Halo. Watch the video and see what knowledge and thinking is required to play a complex game like Halo. Numerous patents have been filed on this invention starting from 2006, called Human Level AI (aka artificial general intelligence).
all comments are welcomed
http://www.humanlevelartificialintelligence.com
