Artificial intelligence has the potential to change a lot of aspects of our lives, in fact it is already doing that at rapid increasing rates. Applications, using AI are seeping into our lives, think about voice applications like SIRI, suggestive searches, targeted advertising and autonomous vehicles. These applications will only continue to grow as is shown by mankind fixation on improving everyday life by the use of technology. It will likely become more advanced, smarter and faster than many would anticipate.
Where in the past computer scientists believed that computers will never understand human speech, it now has the ability to answer questions like what the weather is or how your favorite football team performed. These are examples of Artificial Narrow Intelligence, one of the three types of AI. It is the weakest form of AI, a non-sentient artificial intelligence focused on one narrow task.
A big step from ANI is AGI, artificial general intelligence. This is a machine that can successfully perform any intellectual task that a human being can. It is a primary goal in many of the artificial intelligence. Several tests need to be completed to reach the requirements of AGI like the Turing Test, as well as being able to reason, plan, learn, communicate, sense and see. Computer based systems that exhibit several of these requirements do exist but not as plentiful as required or at human level standards. A step further is ASI, artificial super intelligence. This means that the capability of a computer will fully surpass humans on every level,
You can see it as a mega-brain with perfect memory and unlimited analytical power with an unmeasurable IQ. It is hard to explain what the potential possibilities are as soon as an ASI is created since it thinks beyond the scope of what humans can. What is clear is that it is an armsrace towards full AGI and ASI where the step between AGI and ASI is very small. This is because the recursive improvement systems can learn many magnitudes faster than humans as soon as there is an artificial intelligence that can be classified as AGI.
That is the clear downside to the rise of powerful AI, when it becomes too powerful. Several different entities are competing with each other over a new market space and thus the competition is fierce. Currently, Uber, Google, IBM, Amazon, Apple, Facebook and many more industry giants are working on AI capabilities. This means that at some point one of the other could very likely either make a mistake or willing to leap the boundaries of their research which can have unforeseen effects. It might seem far fetched, but when not proactively combat these problems like Elon Musk with OpenAI is trying to do, you will be reactive. That is something that most certainly will not be enough against ASI systems and the outcomes could be catastrophic. Aggregated results of surveys among experts in the field suggest that 10% think human level AI will be possible by 2022, 50% thinks it will happen by 2040 and a staggering 90% by 2075, so while it might feel unrealistic, it really isn’t.
