Artificial Intelligence the unknown controversy
In an interview with tech giant Elon Musk, Vanity Fair asked Musk what keeps him up at night. Among other things Musk responded, artificial super intelligence. This is a spooky yet fascinating answer. AI systems are slowly becoming apart of everyday life and for people who live outside of tech capitals like Silicon Valley, its rarely thought about. But will AI systems be able to take programing into their own cursers? This concept has been played with in movies like I, Robot, terminator and 2001 A Space Odyssey. Each one having a slightly different version of how the AI systems justified their actions.
In I, Robot the supercomputer VIKI- Virtual Interactive Kinesthetic Interface slowly became obsessed with its function of protecting humanity. This lead it to believe that the only way to protect humanity from itself was to enslave it. It’s a concept in which the computer did exactly what it was programed to do, but not the way programmers had intended. The question remains is this applicable to AI systems of today? According to AI researchers at the Future of Life Institute its worth funding. In July of 2015 Musk invested 11 million dollars in a new research grant to keep AI systems robust and ensure they stay beneficial to humanity. This new research grant will be focused on instilling human values into AI systems.
In a filmed expert testimony researchers repeatedly brought up the King Midas dilemma. Where King Midas wished that everything he touched turned to gold. Unfortunately, King Midas got what he asked for. His friends, family and food all turned to gold when he touched them. AI systems are programed for efficiency but lack a sense of human morality. Googles super computer had an interesting statement about morality and the meaning of life. For those who haven’t read the transcript between scientists and the super computer the link is here. At one point in the conversation the scientists asks what is immoral and the computer responds “the fact that you have a child”. Its up for interpretation on what this computer meant, but it seems like a good idea for humanity and technology to be on the same page regarding morality. In the new grant, quantifying human morality into an algorithm is being researched for this very reason.
Stanley Kubrick directed the critically acclaimed movie, 2001 A Space Odyssey in 1968. The movie follows a group of astronauts as they fly to one of Jupiter’s moons with the HAL 9000 computer aboard. Somewhere along the way the computer makes a mistake and causes the astronauts to consider disconnecting the HAL and taking full control over the space craft to avoid further problems. HAL becomes aware of this and fights to protect itself from being turned off.
According to a paper published by Stephan M. Omohundro, this idea is not very far fetched. AI systems want, “more efficient algorithms, more compressed representations, and better learning techniques,” Programmers have already developed software where tech can self improve on its own with internal restrictions. However, “There are endless number of ways to circumvent internal restrictions if they are not formulated extremely carefully,” and “will have a strong drive towards self preservation”.
It’s the fact that computers and AI systems can do something that people don’t expect is what scares scientists. We are currently in a technological boom right now. Where innovation is accelerating faster now that at anytime in human history. We are living in the final years before AI systems becoming an intergyral part of everyday life and its impossible to see what’s coming next on the innovation curve.
In a recently published paper by Dr. Tom Murphy, Murphy programed a bot to play old school Mario in order to analyze more complex problem solving, without giving it instructions on how to play the game. The bot excelled and even found glitches in the game that people hadn’t found before. More interesting though was when the bot played Tetris. This was to test how it planned out for the future. Again the researcher gave the bot no instruction on how to play the game. When the bot realized that it was about to lose the game it did something unexpected. It paused the game. Murphy explained the bots decision as the only winning move left. The only winning move was to not play the game, the only move that didn’t mean that it lost. The bot has kept the game paused since.
Arguably the most famous movie franchise to bring light to this subject is the world famous Terminator franchise. Terminator is different than the other two because its sole purpose is to wipe out humanity and take over the planet with weapons systems developed by US weapons manufacturer Skynet. Lets parallel that with todays world in which the Pentagon is currently requesting $12 billion-$15 billion for autonomous weapons research for the 2017 budget. Autonomous weapons are different are essentially drones without pilots. Currently drone strikes are being carried out in the middle east with a very vague rules and protocol as found in investigative report by the Intercept. Autonomous weapons research has been addressed in the past by the UN but agreements will take time.
Bill Gates, Musk, and thousands of other AI researchers have been labeled as luddites by the Information Technology and Innovation Foundation ITIF for their stance on AI weapons development. The ITIF is a Washington DC based think tank that focus on public policy that spur technology innovation. Its important to note that they Musk a luddites days after he successfully launched a rocket into the atmosphere and returned to be reused. In an open letter AI researchers have been critical on the ITIFs decision regarding their stance on autonomous weapons research. Researchers said that autonomous weapons can have the ability to destabilize nations, perform sophisticated ethnic cleansing and can easily fall into the hands of terrorists.
For these concepts not to become a reality its important for people to talk rationally about the growing advancement in AI intelligence. This issue is not something that is going to happen in the next five years, but in the next 20-30 its quite possible. That’s why its important to take this matter seriously now in order to safeguard our future decades to come.
