AI : World Domination or Advancement?
Before you start this read here is some data I referenced:
In 1951 something new was created. Something that had the ability to change our vision on life and speed up our life. That thing is Artificial Intelligence. A man named Dean Edmonds was the first person who built a neural net machine. A neural net machine is a part of computer science that has to do with building algorithms and teaching a machine to learn itself. This is what Artificial Intelligence (AI) was created to do, to learn how to do our jobs better. With Artificial Intelligence getting high recognition as of 2017, many people think we should stop trying to teach a program to learn by itself. As a student learning computer science AI strongly interests me and most of all, the fact that robots can soon become like humans scares and interests me as well. I disagree with most people that say if we code a program to learn it self, it eventually will take over and turn on humans. This philosophy people have on AI machines raises a concern to all, should we stop teaching robots to learn?
When the Stochastic Neural Analog Reinforcement Calculator was created in 1951 AI was born. This was a very simplified version of AI as the machine would calculate a probability from a signal and go through 40 machines through a chain and check if the clutch in the machine was engaged. This was a very complicated task for a little idea, and at this point most people did not know about AI and there was no fear against it. Moving along to 2017, Googles subsidiary called DeepMind created a Go-playing software, AlphaGo Zero. Go is a traditional Chinese game that was invented in ancient China more than 2,500 years ago, and is a game that requires all skill and is an extremely difficult game to play. Some professional games exceed 16 hours and are played in sessions spread over two days because of the difficulty of the game. AlphaGo Zero is supposed to be better than the first program they made that beat the game’s world champion earlier in 2017, but more importantly, it is entirely self-taught. This is an advancement of AI, and this is one region people fear. Having a program can learn and teach itself a game completely by itself shows independence, but not from a human, from a robot. It learns by playing the game repeatedly and learning new strategies, and now, has no constraints of human knowledge. This makes the program can learn itself from the first principles, from a blank slate. Since this program was not programmed to understand Go specifically, but more to learn Go itself, it can be applied in many other fields to discover information. Another recent AI robot which strikes fear into humans is the Sophia robot by Hanson Robotics. This is a social robot that uses artificial intelligence to have conversations and can recognize humans. This robot understands and over time learns and increases her intelligence which is one idea that scares humans.
I myself believe that we should allow robots to learn, as there is a cap on human intelligence and if we want to succeed and solve problems we need much more processing speed than our brains can handle. These robots can help speed up solving our world problems. For example, AI has already been helpful to our lives. For example,
This information from Cornell University Library shows that AI has been used to help benefit the medical field. Without machine learning and allowing the program to solve this problem, it would take much longer to do. This is an example of contained AI, this Artificial Intelligence is teaching itself to learn about retinal images, however does not have the ability to learn about other items, like world domination?
Another reason letting AI learn for itself is a good idea is the AlphaGo Zero program. The starting of this program was like the medical program designed by Cornell University. It was a program that is searching for information/solutions on how to solve a problem. The downside of Cornell Universities application is the fact that it cannot be applied anywhere else without major changes to the application. AlphaGo Zeros machine learning technique started off poorly, but still beat the best human in the Go game. With the second version being different and allowing the machine to learn itself, it allowed usage of the application for other reasons. Since Zero was not defined to only understand Go, it could easily be reprogrammed to discover information in other fields, this means we could apply Zeros code to quantum chemistry, particle physics, and multiple other fields. This is a huge advantage to allowing programs to teach itself what to do, as now that they could learn, they can help solve many problems that take us years to figure out.
And my last point to why we should continue to code and improve AI techniques and allow these machines to learn itself is the example of Sophia. This example isn’t very practically, but it does show how we can constrain our advancements. Allowing a program to teach itself more information is scary, but the amount of information needed to allow a program to be smarter than every human in the world is massive, and even that would take a lot of memory space which of course could become a hassle with AI. Sophia is a great example of a robot that could take over the world. This is because it has human intelligence and is constantly learning more and more about life itself. But Artificial Intelligence is so far from independence, and the theory that machines will take over the world shouldn’t be a fear. Sophia shows emotion and shows that humans still have control over machines and can turn them off whenever they want.
Out of 43 people, 27 people thought that robots could take over the world. The most common miscaption is if we teach machines to learn, they will eventually become smarter than humans and take over. I did a survey where this data came from and one person said,
“Artificial intelligence is the next step to elevating our world to a new level. As long as it’s done knowing the fact that the AI could become too smart for its in good”
While another person said, “AI should be something developed to further human innovation & make human lives easier, healthier & better as a whole. that being said I don’t think we should have to depend on the services offered by AI for survival, that poses some major risks to humanity and creativity.” This shows the difference between the arguments. Both are for using AI but the second person said that AI could pose major risks to humanity. This concern of machines taking over is called “AI Takeover”, and isn’t going to be possible for a long time. The reason we should work on improving artificial intelligence is, so we can use more brain power in solving more problems. Without machine power we would fall behind as we aren’t going as fast as we could with machines. An example of this would be Bioinformatic, a Bioinformatic is a field that develops methods and software tools for understanding biological data. Using AI can help this process of finding methods in the biological field.
This quote from PubMed.gov is talking about how using a program, domino, that they created, they were able to learn how to extract information from a broad array of features. This tool of artificial intelligence is helping the medical field where it would take a much longer time for humans to predict a gene type. The advantage of using machines learning, teaching machines to learn themselves, allows predicting gene types more accurate and there is no threat that the machine will come to life and take over.
Machines, robots, and programs all live in our daily human life. Teaching these programs to learn it-self will not only benefit mankind, but will increase learning capabilities and the speed of problem solving. Without machine learning, these artificial intelligence programs could not help us in times when we need it, when we need to predict something that could save the world, or end it. With predictions and data human error is likely to appear, but with a more precise tool and a computer that is learning from itself, the mistakes are less probable. If making sure the earth is safe, that we are solving every problem as fast as they come and saving a ton of lives, letting programs learn themselves will benefit mankind, and save us a ton of time.
Thank you for the read, Artificial Intelligence and Machine Learning is something I love and eventually hope for everyone to believe in the advantages of using them in our discoveries.
Roshan
