Can an AI Build Another AI? |
Posted: February 26, 2020 |
Billions of years ago, the strongest creature with the strongest MIND was born on the earth. Who was that? Of course Humans! Humans born and started inventing infinite things. Tons of things occurred during these past billion years! Everything was damn simple when humans started life. No food, no electricity, no jobs, no taxes, no complicated relationships, and no TECHNOLOGY! Now, you will ask if life was simple in the absence of all these things then what is the point of inventing them? Good question! I said life was SIMPLE but it became SIMPLER with jobs, foods, and electricity. And now, it turned the SIMPLEST with technology like artificial intelligence marketing. The evolution of human intelligence to artificial intelligence We can understand it like this. Remember the time when we used to write letters to communicate. Then, we invented phones, and now smartphones with abundant features. Similarly, we invented computers, then its programs and now technologies like AI and ML that can answer our queries.
How artificial intelligence can create another artificial intelligence? Can you guess it? Yes, you are right! It is possible with the help of genetic algorithms. On what power our brain works? It is 20 megawatts. But, we already created something that runs on more power than our brains. True! These are computers. What computers lack than we humans? They lack the brain architecture as we humans have. So, if we power their code with architecture like the human brain, then they can work with equal efficiency. They will start evolving their code if we:
A successful example of the child AI by Google Google created an AI (artificial intelligence) system that further built its own “child”. The best thing is that the original artificial intelligence provided training to its own “child” to that much level that it is outperforming all human-based AI systems similar to it. I must say that it is a great achievement. But, one question that hits our mind is that will AI invade all areas of human capabilities? Do you think so? Definitely not! AI has a far way to cover to match the conscience of humans. In May 2017, Google announced its AutoML project. The experts say it outperformed so many humans tasks. But, what are the parameters of quality? Again the machines depend on humans to evaluate the quality of their work. They can perform tasks but quality assurance will be provided by people. With feedback, experts inform the controller to improvise the performance for upcoming rounds. They repeat the process (keep testing it) by the time they start receiving accurate results. Wrapping up Hence, you got to know by now that an AI can surely create another AI. But, it cannot kill humans existence from each sphere. It is intelligent only when we provide it accurate facts. Humans can be blamed for their errors but you cannot blame AI for a software/hardware crash. Therefore, both can work together with the power of each other. If you feel I missed any point, you can share it in the comment section below. I will soon post my next blog. Stay tuned!
|
||||||||||||||||||
|