	In Stanfords report of a long-term research of Artificial Intelligence (AI) named The One Hundred Year Study on Artificial Intelligence, they cite Nils J. Nilssons definition of AI. According to Nilsson, intelligence is a quality such that an entity would be able to solve and have foresight to problems if it is intelligent, and artificial intelligence refers to machines that are made intelligent by humans. Authors of the report furthermore quotes Nilssons idea that different kinds and different level of intelligent entities all lie on a spectrum with multiple-dimension. To explain this point of view, they use calculator as example. As they state, a calculator shares not much similarities with nowadays AI, though it is put on the spectrum by broad interpretation (p.12). This suggests that although simple devices such as calculators do not look like what people expect todays AI would be, they are AI. To conclude, authors of the report propose that manmade electronic devices that are able to complete tasks humans once completed should be regarded as AI.
	From my point of view, electronic devices such as calculators that cannot make decisions by themselves could not be regarded as AI. Only devices that have ability to make decisions could be considered as AI. I prefer to interpret the foresight in Nilssons definition of AI as an ability to determine what outcomes every solution could bring and consequently find out relatively the best solution to a problem. Therefore, in my opinion it is not what Stanfords report writes that as long as a device could perform the activity humans once performed, it could be seemed as AI. 
	As the National Association of Credit Management (NACM) discusses in Artificial Intelligence in the Credit Department, calculators are only programed to take job of calculations for humans, and since they lack the ability of making decisions, they could not be regarded as AI. This example suggests that only devices with decision-making abilities could be called AI, which indirectly shows my point of view. 
	Moreover, Fernando Schwartz, a Mathematics Professor at University of Tennessee, states in his article One Data Scientists Thoughts on Highly Intelligent Machines that when doing calculations, humans and calculators would probably adopt the same algorithm and the only distinction between them would be the speed of calculationcalculators could definitely calculate much faster than human brains. Schwartzs opinion again shows a calculators attribute of a tool. Although a calculator could enable humans do calculations more quickly and precisely, it does not do calculation itself and all algorithm it adopts are still programed by humans. As a rule of thumb, tools without decision-making abilities usually would not be considered as AI. 
	In Stanfords report, researchers themselves take a scheduling system of an airport as an example of AI. However, their reason for providing such example is still about a scheduling systems calculation ability. From my point of view, a scheduling system indeed could be regarded as AI, not for its capability of calculation but for its ability of making complex decisions. Different from calculators directly executing what humans expect them to execute, a scheduling system of an airport should know about all possible choices of scheduling every plane and find out the relatively best schedule for all airlines. The system is making decisions for humans, based on algorithm written by humans. Therefore, a scheduling system of an airport could be called AI because of its ability of decision-making, not because of its ability of calculation.
	In conclusion, I think only electronic devices that are able to make decisions for humans could be regarded as AI.
