I think it is very important to note the difference between artificial intelligence and artificial general intelligence. In the public mind, these are two different names for the same goal, but really most people are discussing artificial general intelligence. There are many different kinds of intelligence, just as there are many different kinds of skills. One person can be good at chess, while another studies art. Both of these people would be considered knowledgeable or intelligent in their respective fields. Similarly, a computer that masters chess could be considered intelligent in its own domain, but it is obviously not generally intelligent; it cannot hold conversations or reason philosophically. Artificial general intelligence is the technological equivalent of human intelligence, while artificial intelligence can also come in domain-specific applications. By this logic, the recent trend of specific artificial intelligences, such as AlphaGo and Watson, are indeed forms of artificial intelligence. These are no mere parlor tricks, they show an ability beyond that of human reason, and exhibit emergent behavior in their choice of actions (look at the second game of the recent AlphaGo vs Lee Seedol series).
This of course brings up the question of the Turing test. Alan Turing described a test by which a person has conversations with both a computer and a person, without directly seeing either party. The objective of the test is to determine which party is the person and which is the computer. When the computer is not consistently identified correctly, it can be said to have passed this test. The Turing test does not, in my opinion, indicate any real level of artificial general intelligence. It does however allow an artificial intelligence to show proficiency in human communication. The Chinese room thought experiment does a great job of showing this distinction. Acquiring this ability is an important step in creating useful AIs in the mind of the public, as it will be much simpler to interact with the computer through natural language than hardware devices. This barrier will continue to shrink as we get closer to artificial general intelligence, as users queries and tasks will become less restricted by the intelligence’s abilities.
While there is a mountain of work between the current state of the art and a conscious, thinking AI, I believe that we will reach that point eventually. There has been fascinating research done with the C. Elegans, a worm whose entire neural network has been digitally mapped. Researchers were able to take the design of that network, replicate physical senses with sensors, and by passing that information through the network, create realistic action with a robotic worm. While the gap between that neural network and the human mind looms large, this is a great example of the possibilities that artificial intelligence has to one day create an AGI. However, I am not sure that these approaches will be able to create the more intangible parts of consciousness – that of emotions and desire. At the end of the day, to me that is what describes a true “mind” – not just the ability to reason, but want and feel. In short, the human condition describes a true mind.
kosercseethics 