Artificial intelligence is becoming more and more complex, and playing games with humans has become a very tricky thing. Researcher Arend Hintze explored the future of artificial intelligence and games. As early as the 1980s, a teacher challenged me to write a game called "Tic Tac Toe", which I unfortunately failed. But just a few weeks ago, I explained to one of my computer science students how to use the so-called "great mini algorithm" to complete the "Tic Tac Toe" game program. Finally, we spent about an hour writing it. This app. It is undeniable that my programming skills have been improving over the years, but the progress of computer science has come a long way. What seemed impossible in the past few decades has become extremely easy today.
In 1997, in a six-game chess game, an IBM computer called "Deep Blue" defeated chess master Garry Kasparov, which was shocking. In 2015, Google revealed that its DeepMind system has mastered several video games of the 1980s, including a key winning strategy.
In 2016, Google's AlphaGo system defeated the number one Go player in the fifth game. The exploration of technical systems that can defeat humans in the game continues. At the end of May, AlphaGo will compete against Ke Jie, the best player in the world, at the Go Summit in Wuzhen, China.
As computing technology improves and engineering improves, computers can even defeat humans—such as poker—in games we think we need to rely on human intuition, intelligence, deception, or bluff. I recently saw a video in which volleyball players practiced and slammed against robot- controlled rubber arms in an attempt to stop shooting. I got a lesson: When the machine struggles to win, human efforts are futile. We all want a perfect artificial intelligence system to drive our cars, a tireless system to look for signs of cancer in x-rays, the idea is wonderful. But when it comes to games, humans don't want to lose. Fortunately, artificial intelligence can make the game more interesting and even bring endless happiness.
Today's game designers, who write something more profitable than a blockbuster, illustrates a problem: it makes no sense to build an invincible artificial intelligence system. No one wants to play a game that cannot be won. But people really want to play immersive, complicated and exciting games. Even when it's the best game now, it won't be new after playing for a while. Interesting games will allow players to adjust and react to keep the game fun, perhaps forever. Therefore, when we are designing artificial intelligence systems, we should not look at the fascinating "dark blue" and "alpha" artificial intelligence, but to see the hugely successful multiplayer online games like World of Warcraft. These games are all graphic, but their main attraction is their interactivity.
Most people don't seem to like the extremely complicated logic puzzles like chess and go, but prefer meaningful connections and community activities. The real challenge of these massively multiplayer online games is how to keep the players new every time, rather than whether they can be defeated by intelligence (man-made or non-human). Currently, many game environments allow players to interact fully. The role played in the squad in the dungeon is clear: the warriors are hurt, the healers help them recover from the pain, and the fragile wizards cast magic from a distance. Think of Portal 2, a multiplayer game that is completely focused on robot collaboration, getting them lost in a labyrinth of cognitive tests. Exploring the world together can make you and your friends share common memories. But any changes to these environments or basic layouts must be made by human designers and developers.
In the real world, change occurs naturally, without supervision, design, or human intervention. Players can learn and creatures will adapt. Some creatures will even evolve together and respond to each other's development. (Similar phenomena have also appeared in the arms technology arms race.) Today's computer games lack such complexity. For this reason, I believe that the development of artificial intelligence capable of playing modern games does not have a substantial advancement in artificial intelligence research. A game worth playing should be unpredictable because it adapts, maintains a permanent novelty, and it is created by playing games. Future games are yet to be developed. The role of the game should not just be a reaction; they need to explore, see weaknesses, and learn to cooperate.
We know that Darwin's theory of evolution and his knowledge are the driving factors for all the novelties on earth. This may also be the reason for the change in the virtual environment. Evolution can explain the creation of natural wisdom. Shouldn't we turn the development of artificial intelligence into evolutionary artificial intelligence? Several laboratories, including myself and my colleague Christoph Adami, are studying "neural evolution."
In computers, we simulate complex environments such as road networks or biological ecosystems. We create virtual creatures and challenge them to evolve thousands of virtual descendants. Evolution itself is the best driving force, or arguably the best organism to adapt to the environment, they are living creatures. Today's AlphaGo is in the process of learning by continually playing games with itself and analyzing the game records of the top Go champions. But it won't learn like us and experience unsupervised experiments. And it can't adapt to a specific opponent: for computer players, the best action is to determine, regardless of the opponent's style. The program learned from experience is the next step in artificial intelligence. They make computer games more interesting, allowing robots to not only play a bigger role in the real world, but also to adapt the robot to its operation.
Dc Power Cable,Dc Power Extension Cable,Dc Power Cable Connectors,Dc Cable Connector
Dongguan Swan Electronic Technology Co., Ltd , https://www.swanconnector.com