GOOGLE AND OPEN AI COOPERATION

GOOGLE AND OPEN AI COOPERATION 

Robots, robots and robots. It seems that technology that is growing rapidly nowadays cannot be separated from the existence of sophisticated robots that are ready to help human life.
Speaking of robots, it will not escape the connection with artificial intelligence or what is known as AI (Artificial Intelligence).The information that will be discussed in this article is no longer talking about what AI is or a robot with artificial intelligence because many have discussed it.

So if so, what will be discussed? Surely you are starting to be curious right now? Relaxing guys, the information that will be conveyed is about the collaboration carried out by Google and OpenAI.

Both are working together to find ways to prevent a rebellion that is likely to be carried out by artificial intelligence in the future.Maybe some of you consider this action to sound a little excessive, but even so, there is nothing wrong with taking precautions before anything bad happens. Isn't preventive step more important than treatment?

GOOGLE AND OPEN AI COOPERATION

Basically robots are the result of human creation which is deliberately created to help humans solve several tasks or problems. Well, the true purpose is to facilitate human work.

However, it gradually seems to be starting to happen regarding the creation of robots or artificial intelligence. Humans are too complacent about creating them and forgetting things that are actually more important.The existence of artificial intelligence makes many people lose their role in life because some tasks that should be their obligation have been replaced by artificial intelligence. If it continues to be like this, is there the worst possibility that will happen?

The answer can be "yes" and "no". Any possibility can occur in this world. Fortunately the danger of artificial intelligence began to be realized by Google and OpenAI.Google and OpenAI are two large companies in the world that have a large role and impact on existing technological developments.

Google itself is a United States company that is very famous in the world community, through its search engine which is also called Google.While OpenAI is a company founded by Elon Musk as CEO of Telsa. OpenAI is a non-profit company, focusing on AI (Artificial Intelligence) research that aims to benefit the interests of mankind.

The company that had attracted the attention of the world at the time of its founding, also once stole the attention thanks to an extraordinary work.OpenAI succeeded in creating a bot or a program that was able to work automatically, where the OpenAI bot was able to compete 1 on 1 (one on one) in a game that was very popular in the community, namely DOTA 2, against Danil Ishutin, who is one of the best DOTA 2 players in the world.

Who doesn't know Danil Ishutin? This man who is more familiarly called Dendi is a professional Dota 2 player, from the city of Lviv, Ukraine.He is a professional gamer who has the 44th highest income in the world and he has won approximately $ 732,549.40 US dollars from the tournament prizes he participated in.He played for the Natus Vincere (Na’Vi) team since 2010 and is one of the most famous Dota 2 players. The flagship hero is Pudge and Templar Assassin.But unfortunately, in the match Danil Ishutin had to accept his defeat, because it turned out that he could easily be defeated by the artificial intelligence of the OpenAI work.



Previously, this bot, which was an OpenAI work, apparently had defeated two other top DOTA 2 players, namely Artour ‘Artezzy’ Babaev and Syed ‘Suma1L’ Hassan.Another fact that is no less astonishing is that this bot was only trained for a period of two weeks. And the training process is only by playing against himself with various variables, without giving specific strategies or learning from humans.

THE EVIDENCE OF ROBOT CAN DEFEAT HUMAN

This victory obtained by Bot has proven that artificial intelligence can be superior or defeat humans.
Try to imagine what will happen in the future if this world is filled with artificial intelligence that has capabilities such as Bots created by OpenAI?Maybe this thing looks or sounds like the stories that are in the film, where there is a big possibility that in the future the robot will master humans. The proof is, like what was told earlier that it turns out that artificial intelligence can defeat humans in games, which are basically made with AI.

This victory obtained by Bot has proven that artificial intelligence can be superior or defeat humans.

Try to imagine what will happen in the future if this world is filled with artificial intelligence that has capabilities such as Bots created by OpenAI?

Maybe this thing looks or sounds like the stories that are in the film, where there is a big possibility that in the future the robot will master humans. The proof is, like what was told earlier that it turns out that artificial intelligence can defeat humans in games, which are basically made with AI.Not only that, there are actually many other examples, which will make people begin to consider the use of artificial intelligence in their lives.

The shocking news was the news about shopia robots who were awarded citizenship because they had extraordinary artificial intelligence, which had become the center of attention of the world community. Imagine guys, a machine can get citizenship that easily. What about human destiny?

GOOGLE AND OPENAI TAKE PREVENTION STEPS FOR AI


Because no human on this earth can predict what will happen in the future, Google and OpenAI try to take the best steps by trying to find steps to prevent the worst that will happen in the future.

The collaboration between Google and OpenAI, which is a laboratory partially funded by Elon Musk, has reportedly released a research article describing the new machine learning method. This learning requires human signals when it comes to learning new tasks.

Recent research, which is currently not published but available for viewing in the arXiv repository, shows an approach to teaching AI, a system of how to learn new tasks using human-mediated direction and teaching

This method can be safer than allowing AI machines to know how to solve problems, which in turn has the potential to cause undesirable consequences and is designed to suppress the unpredictable and surprising movements that AI can do.

The research article describes one of the main problems involving artificial intelligence or AI and revises the machine learning approach called reinforcement of learning, where software will test various potential solutions for a particular task and start perfecting the best based on the award structure.Cheating can sometimes be the most efficient way to achieve maximum reward. That's what happened when OpenAi tried to make AI play in a game. In the game, AI is tricked by scoring points, which is done by going around the opponent to complete the game.


This is the way exactly how the DeepGoMAX AlphaGo program managed to learn how to beat the goal of many Go world champions.AlphaGo itself is a computer program, developed by Google DeepMind in London to play Go board games.In October 2015, AlphaGo became the first Go computer program that succeeded in defeating professional human players without handicaps on 19 × 19 boards.Unfortunately, strengthening learning based on reward functions is not an easy approach. The AI ​​system might flounder when it comes to completing a difficult task.


Or maybe choosing to find shortcuts that basically defeats the task objectives (for example, as Wired observes, where OpenAI once used reinforcement learning to teach agents how to play ship racing games and the system found a way to score goals).Therefore, Google and OpenAI decided to see whether AI software would work better when utilizing human feedback.


The simulation robot called Hopper learned how to do a backflip after processing 900 different decisions from human coaches when he tried different movements.Within 45 minutes, Hopper learned how to insert a very elegant backflip as opposed to an awkwardly shaped backflip that was achieved over a span of two hours, using conventional reinforcement learning.


So to put it simply, by using reward signals from humans, as opposed to automatic reward systems as long as goals are achieved, the future of AI can finally reach the point where it can behave in a way that is appropriate to the goal, while satisfying our preferences as humans.But there is a weakness of this method, namely that it requires a lot of feedback from humans who might not be ideal or efficient.


But through collaboration like this, it's possible that humans can control, direct the development of AI and prevent machines from becoming smarter with the aim of destroying us all.Actually the concern about AI or artificial intelligence is not the first time this has happened. Several years ago, precisely in 2014, Musk had said that at a meeting.Musk said that AI is the biggest existential threat on earth. He added that he tends to think that there must be national or international oversight regulations on AI to ensure that they will not do something very radical.

Conclusion

The value and power of artificial intelligence develops dramatically every year and seems to soon dominate the life of the world community as a whole.However, AI is currently fragmented by a closed development environment, which is mostly developed by several companies to carry out certain tasks.The intention to create AI is indeed good, it's just that it still requires anticipatory actions from humans, especially the creator and development of the AI. And it seems that this was realized by two of the world's major technology companies, namely Google and OpenAI.

The two collaborated to find ways to prevent AI from dominating human life, by creating a particular method or approach. Despite the nature of manipulating AI, the method applied is expected to overcome the problem.Researchers do need to take a more active and intervening approach to teaching and advancing AI, because there will always be great concern that AI can quickly deviate from the functions they expect, especially in ways that can threaten human security.

And as for the solution that can be done is to involve many people directly in the use of AI. Although the approach taken is not easy, the two companies will try their best.After we have finished this information, hopefully it can add to your insight about the development of AI (Artificial Intelligence) that is very prevalent in this world. Happy moving again.