How Dangerous is Artificial Intelligence?
Many times in history, the human imagination has given birth to former science-fiction fantasies. From pocket computers, to self-driving cars, space tourism, virtual reality, and now, artificial intelligence, humanity has blurred the lines of fantasy and fiction through innovation and curiosity.
Artificial intelligence (AI) is now a very real prospect that companies are focusing on. For those of you who are still new to this concept, artificial intelligence is a field of science which focuses on how hardware and software components of a machine can exhibit intelligent behavior. Instead of being fed information from the user himself, they learn over the course of time and become more intelligent.
Currently, many companies are working on AI projects including Microsoft, Google, Facebook and Minecraft. South Korea also has some high profile AI projects in development. And these are just the companies that have made this official. Let’s not forget there might also be companies that are secretly working on AI projects.
Why should we be worried about Artificial Intelligence?
Today, artificial intelligence, which was once thought to live purely in the human imagination, is very real. In recent months, artificial intelligence has emerged as a topic of great interest for many leaders in the tech industry.
First there was Stephen Hawking, then Elon Musk, and most recently Bill Gates. All of these smart people have suggested that artificial intelligence is something to be watched carefully. When Tesla CEO, Elon Musk was asked about artificial intelligence, he said it was like “summoning a demon” who shouldn’t be called unless you can control it.
When Stephen Hawking was asked this same question, he cautioned the public by saying that any further advancement to artificial intelligence could be a fatal mistake. He mentioned that AI has the power to re-design itself and take off on its own whereas humans have slow biological evolution, and they wouldn’t be able to compete.
According to Bill Gates, AI devices will be fine initially, but as they start learning more and more from us, and about us, they will get more powerful and intelligent than the humankind.
The risks of Artificial Intelligence
There are four classes of risk posed by artificial intelligence:
- Malicious hostile risk – The only two feasible scenarios by which a maliciously hostile AI might be possible are if it is deliberately programmed to be hostile (e.g. by a military or a terrorist group), or if humanity’s existence or behavior is actively and deliberately contradicting one of the AI’s goals so effectively that the only way to achieve that goal is to wage war with humanity until either humanity’s will or capability to resist is destroyed.
- Apathetic risk – There is effectively no risk of apathetic danger from an AI with a friendliness super-goal but it is almost unavoidable from an AI without. An apathetic AI is dangerous simply because it does not take human safety into account, as all humans naturally do. For example, without friendliness goals, an AI in charge of dusting crops with pesticide will dust a field even if it knows that the farmer is standing in the field inspecting his plants at that moment.
- Accidental risk – An artificial intelligence working with incomplete data is capable of misjudging, just like a human. Mistakes of this sort are almost inevitable since it is impossible to know everything there is to know about the world, but they are also the least dangerous of the four risks. Since AIs can learn from their experience, the occurrence of accidents actually decreases the chance that the mistake will happen again, improving the AI and making it safer.
- Unknowable risk – The real danger of well designed artificial intelligence is in its ability to reprogram and upgrade itself. Any AI capable of self-improvement is likely to eventually surpass the constraints of human intelligence. Once an artificial intelligence exists which is smarter than any human it will be literally impossible for any human to fully understand it. Such an AI is also likely to continue improving itself at an exponential rate, making it increasingly impossible to comprehend or predict. Also, at some point, the AI may discover laws of causality or logic far beyond the comprehension of human minds, and the possibilities of what it can do become literally infinite.
While there are a lot of assumptions about artificial intelligence being dangerous, we have to remember that these are just assumptions and not clear facts. Humans have always been doubtful about new technologies, and there was a time when we were also hesitant about cell phones. After all, it’s about how we create artificial intelligence and how we keep it under control.
Photo source: http://www.freedigitalphotos.net/