TinWiki: Artificial intelligence

Artificial intelligence also known as AI is the principle of creating intelligence in machines. A famous quote that is the definition of AI is "the study and design of intelligent agents." The idea is that a machine with artificial intelligence is capable of making choices, movement around obstacles, solving problems and thinking about others reactions and issues. A being with artificial intelligence must be able to fit certain criteria, its must complete the following...

• Reasoning
  
• Knowledge
  
• Planning
  
• Learning
  
• Communication
  
• Perception
  
• The ability to move and manipulate objects.
  

To achieve AI some people think that we must not make the machine able to choose options but instead program it to have all cases already in its components so that it can just pick and choose, it sees a problem -- goes through its library of reactions and then chooses the correct response, which in a sense is and is not AI.

Issues in A.I.

At the current time, A.I. is very difficult to create. Machines are unable to understand things such as emotion, reason and change. All they operate on is logic and data. Thus, a machine could learn to talk, but never understand the subtle nuances of the words, or what we would call 'semantics'.

The solution would be to create a machine that can learn from it's environment and thus grow to understand such things as a child does, but the complexity of such a system would be enormous, and will likely be beyond our means for decades.

The worry of A.I. for many is that robots that can learn will grow to be of equal or better intelligence than humans, which combined with the fact they can build themselves to be far superior physically, could pose a threat to humans. This is shown in many popular films such as The Terminator and The Matrix, or perhaps a more likely situation in: I-Robot.






External Links

What is AI -- by John McCarthy, one of the worlds leading people on this matter.

reply to post by WikiEditor


This should have things about the singularity as well.

Humanity+

Open Cog

People need to know that the likely strong AI will rise from the internet and be a combination of the vast amounts of humanity.

I think it would be appropriate to have a section about the hypothesized need for any AI machine to also have a analogous artificial emotion component, coupled to a feedback loop, which would drive the machine's selection processes like those of any intelligent creature. "Intelligence," of course, is only measured against what we understand to be intelligence, and that is judged essentially by referencing ourselves. That would be difficult if not impossible unless the machine had a simple, synthetic emotion core working with the memory and perceptual components, which the machine would use to make choices based on individual, learned preferences and desires.

P.S. -- Creating artificial emotion isn't all that difficult. Simple programs such as those found in small "artificial pets" (Tamagotchi) popular with the Japanese will do to start. All it has to do is count instances of a particular thing happening, and relate it to a fading scale that the machine is required to act on the lower it gets. Tomagotchis cry when they need attention. You push a button, and it settles down, just like a human baby. Create a program called, "Positive Human Interaction," and every time the machine is petted, or called by a nice name, or sees a smile on the face of a human looking at it, it gains points. If none of that happens, the points diminish over time, creating an artificial "pain," or desire. Program the machine to always keep the points above a certain level, and it could eventually want to get more points, to make its creators proud (get more PHI).

The emotions for the machine would start out simple and strong, like those found in babies, but by adding more emotions, and combining them they would create more complicated feedback loops. Eventually, by allowing the machine itself to make slight adjustments to the points it needs in various categories, it would no longer make simple decisions based solely on the mathematics. It could learn to "suffer" a little to get something it wants. Work harder to get more PHI, or love, or a "promotion," or independence.

The programs already exist. All they need to do is start simple, and slowly be added, just like what happens when a person grows and learns. The feedback system would be the most complicated to balance. A lot of people don't even have a good feedback system. But in the end, the built machine will have intelligence that makes sense to us in a human context, not just how many bits of information can be stored or processed at what speed.