By Tanish Pradhan
Entrepreneur and tech whiz Elon Musk, at the recent governor’s meeting, voiced his concerns regarding the impact of artificial intelligence and the need for regulations on the same. He said, “AI represents a fundamental risk to human civilization.” Artificial General Intelligence is the intelligence of a machine which would enable it to perform any cognitive task that a human could. The threat exists in the possibility that AI could surpass general human intelligence and become super intelligent. Here we take a look at some of the dangers associated with AI and automation while we ponder why great minds like Mr Musk, Bill Gates and Stephen Hawking are crying doom.
How AI could surpass us
If machines do attain superintelligence, it provides them with a virtual brain which far surpasses the cognitive capabilities of even the smartest man on earth. This isn’t just hypothetical. Recent advances in AI research are beginning to show how this is a probable outcome.
It is considered much more difficult for a computer to beat a human player at the Chinese board game, Go due to the much greater number of possible plays. In 2016, AlphaGo, a narrow AI computer developed by Google DeepMind beat the world’s top ranked professional Go player in a three-game match. It also won 60 straight online matches against high ranked professional human players. This goes to prove that AI can already far surpass us in strategic thought.
It’s connection to the internet ensures it immunity from physical dangers and provides it almost infinite longevity. It also provides it with more speed and flexibility in networking than we humans could ever hope to achieve.
With developments in robotics and automation, we provide machines with bodies that exceed ours in strength, durability, and repairability. They are also far more adept at learning physical activities without the need to be taught. This was seen recently when a computer sim taught itself to walk upright within hours.
How things could go wrong
The concern of AI developing a sort of malignant consciousness which chooses to harm humanity has often been the theme of multiple forms of fiction and entertainment. Experts, however, dismiss these concerns, highlighting that they are primarily worried about its ability to perform high-quality decisions. Here, the path taken by the machine to meet the specified goal may not always align perfectly with human values, which are at their best, flexible. The argument is that if the restrictions a machine is subject to are not well thought out, it might lead to a genie in the lamp like scenario, where the desired outcome is lost and the actual outcome is far from desirable. A common example is, a deep intelligence AI with the goal to maximize stock profits would heavily invest in defence stocks. It would then fake news, emails and press releases to strain diplomatic relations and essentially try to cause a war since its only goal is to reap economic benefits. If the AI were smart enough, it could achieve this goal with ease.
Another possibility that many worry about is the idea of an intelligence explosion. This is the possibility that a machine which has surpassed human intelligence would be able to subsequently self-improve by designing superior machines. It would be able to learn and grow at a rate that our brains just would not be able to keep up with. This would make the machines infinitely harder to control. Goal inputs would be resisted and the outcomes would evade our understanding. A greater issue arises when machines perfect the art of deception. Marketing and advertising tactics often focus on ways to use the consumer’s psyche to sway him/her to buy a particular product. Once this is learned by AI, it can be used seamlessly to deceive, manipulate and sway humans into taking decisions which benefit the AI’s objective.
Tipping the scales of power
A major concern is the amount of power we have given machines. Predictions say that 7 out of 10 individuals worldwide would own smartphones by 2020. An AI that was to hack into your smartphone would have access to your microphone, your location, and your camera. It would be able to mine and access all sorts of data gathered on the device. It could even use a person’s phone to understand, evoke and manipulate their emotions using simple cues such as fake text messages or functioning glitches.
With driverless cars being the new direction of the automobile industry, we give machines control over transportation as well. Elon Musk predicts that in 10 years all cars will be driving themselves. Even the defence sector is taking a turn towards automation with funds being poured into automatic drones and robotic soldiers across the world. Little or no attention is paid to accountability in the event of disaster and regulations are close to non-existent.
The manufacturing sector is already going completely automatic. The heavy use of robotics in production lines and the use of computerized machines as compared to traditional manual machinery is already prevalent and growing. This gives machines access to entire factories completely linked by digital networks. A small kink in the production line could impair the quality of or even completely halt the supply of a product. Modifications to the quality of a product could sway customers in favour of a competing product, disturbing the market. Access to manufacturing facilities and automation in the energy sector could mean giving the machines the ability to design, create and replicate cutting edge robots that can perform virtually any task.
The biggest concern
The most pressing and immediate concern, however, that most people have towards AI and automation is their impact on the job market. With the ability to analyse data and perform tasks much more effectively and efficiently than humans can, AI could essentially put a majority of the world’s population out of their jobs. Many argue that this job displacement will only be short term with a greater benefit to society and workers due to increased productivity and better quality of work. They suggest that superior cognitive work will remain with the human workforce. But advances in AI are starting to prove that even work that requires thought and analysis may be performed better and at a cheaper rate by an AI. This could do two things. It could either mean that rampant unemployment and the subsequent psychological distress could lead to a dire situation. On the other hand, it could mean individuals have to struggle less to makes ends meet and can focus instead on pleasure based work.
The economic benefits attained by the rise of AI could be used to enforce a universal basic income. It means that every person irrespective of their income would regularly receive a certain unconditional sum of money from their respective government. This concept is supported by many such as Stephen Hawking, Elon Musk, and Facebook co-founder, Chris Hughes.
Are we on the right side of history?
The benefits AI and automation could bring to our lives are innumerable. It would give us advanced, life-saving medical technology which could predict and detect illnesses far better than human doctors can. It could cause a boom in manufacturing and full capacity energy utilization. It would also mean the eradication of the millions of transport related accidents worldwide. It would cause advancement in research and development on an unprecedented scale. Some would say we are inviting our “Computer Overlords”, while others would argue that we are making the world a better place. But is the tangible benefit really worth the existential risk we are taking? Without the proper regulation, these rapid competitive developments in AI are akin to flying blind. Are we, as Elon Musk said, “Summoning the demon?”
Featured Image Source: Pixabay
Stay updated with all the insights.
Navigate news, 1 email day.
Subscribe to Qrius