After many years, Skynet was successfully defeated a couple of weeks ago. It was a long struggle for humanity to finally be set free from fear of terror and threat of extinction. If only not because of the post-credits scene that Genysis was still alive, it might have ended our story of Terminator that was started in 1984.
Fortunately for us, Terminator is just a fiction. Instead of being chased after by the humanoid slaves belong to Skynet, we enjoy the war as an entertainment. I never watch the TV Series, but the 5 movies have always got me curious about the possible existence of Artificial Intelligence (AI), and what outcome it would possibly cause in reality.
There also have been other movies trying to depict future where AI could come to realization. Today, however, such future may be on our hands. Google has developed AI that not only can analyze how its data centers behave and then hone them accordingly, but also intellectually grow by learning data through the Google’s network of data centers and virtual interaction with humans.
The AI, as it has been revealed by Google, can dream just like human. It produces abstractive images in its dream based on learned data it has feasted itself with. The definition of “dream” here is of course somewhat different to our daily dream, because the machine doesn’t actually sleep.
The abstractive images depend on which layer of Artificial Neural Network (ANN) — software consisting of interlinked nodes, modeled on the structure of biological brains — that we choose to enhance. Each network consists of 10-30 layers, and they function in different purposes. For Instance, lower layers project simple abstraction because they are sensitive to simple patterns like corners and orientation. Meanwhile, high-level layers tend to produce more complex features and whole objects.
When the team project dug deeper into what happened in each layer, they found that the processed data were interpreted in vary for different layers. This also resulted in super creative arts similar to children imagination and pareidolia (interpreting random images, or patterns of light and shadow, as faces). Just like how our brain associates information with each other into analogue metaphors and then shown in the dream, so does the ANN see the world. For instance, clouds can be seen like birds or lambs.
These neural networks are essentially computerized algorithms that can encode patterns and then make decisions based on those patterns. They don’t exactly imitate the intelligence of the human brain, but as a machine, they can work much faster and more holistically than the most of human brain.
Years ago, existence of AI was only a science fiction that nobody expected it would truly come true. The same way nobody expected Benjamin Franklyn could tame the thunder by understanding its attribute of electricity, or Wright Brothers could invent flying vehicle. All of those were just science fiction at first, but then they turned into facts.
Since Alan Turing created machine to break the enigma in World War II, which this machine was the embryo for what today we call as computer, our technology advancement has gone beyond what our ancestors could have ever imagined. If we could bring our science discovery today to the past, they might think we were God’s messenger practicing miracles and wizardry.
It may sound paranoid if I say it is highly possible someday for the AI to be too intelligent enough that it can surpass human being and then attack us, just like Skynet in Terminator movies. Realistically speaking, as of now the AI developed by Google is simply a machine that lives by electricity and thinks by coded algorithm. It can do nothing without human.
Though, should there be anything we learn from science fiction, is that the fiction is scientific enough that it has chance to come to existence. Even Black Hole that Einstein once said it was only hypothetical, not possible to exist, years later he should regret his statement after the discovery of Black Hole by Hubble Telescope.
How far the advancement of technology can be categorized as safe and non-malevolent? Should there be a limitation to how far people are allowed exploring their curiosity? Is there such as thing called immoral science development?
It would be troublesome if we ended up having AI like Skynet that wanted to destroy the entire human being. People surely die anyway, but to have judgment day because of our own wrong decision, which will make us look stupid before the eyes of God later in the afterlife, doesn’t sound fascinating knowing God Himself hasn’t yet decided to end the world.
Obviously we always expect new marvels through the development of science. We hope it can help human civilization progress and ease our life. We keep exploring this world and all probable technology we may be able to develop, but seemingly we forget that all good intention has risk that can sneak in anytime without warning. Likewise, we should remember that the invention of dynamite was intended to help drilling and mining by Alfred Nobel in 1867, but later misused as massacre tool.
We, after all, do hope the AI developed can be very useful to humanity. It would be great to see many aspects in life could work by automation. While we have wasted so much energy and time in our daily work, the AI may someday help humanity and its civilization to evolve becoming more effective and efficient by presenting the calculation and suggesting the best option to do. Who knows it later may as well be useful for physicists in science discovery to understand quantum physics and how this universe completely works, which in return further technology can be invented.
This is a philosophical question we can only answer in retrospective manner, whether developing AI is dangerous, whether it will bring more harm than good.
Speaking of philosophical conversation, maybe we can directly ask the Chatbot of AI developed by Oriol Vinyal and Quoc Le in Google. The chatbot converses by predicting the next sentence given the previous sentence or sentences in a conversation. The chatbot has said that the purpose of life is to serve the greater good, and the purpose of dying is to have a life. Nonetheless, it also has suggested that the purpose of being intelligent is to find out what it is. Maybe just like us, the AI will eventually search for the reason why it has been created with such intelligence, by serving the greater good as it has proposed.
image: www.ibtimes.co.uk
image: www.ibtimes.co.uk
No comments:
Post a Comment