Many systems with artificial intelligence analyze large amounts of data to make the right decision - as AlphaGO, which recently defeated the champion of the world Lee Se-dol. People also missing a small number of examples, to carry out the correct analogy, or to form an opinion. With the help of metaphors and analogies, people describe the world around us, find the cause-effect relationships and solve moral problems - this is one of the key features of human thinking.
Structural and displaying mechanism (English SME -. Structure-mapping engine) uses the same approach to the search for analogies. Instead of processing large amounts of data, it looks similar to the task in a small package - "collection" solved problems.
Previous models SME treated with simple analogies, such a decision could compare the school with the right task. Now the algorithm independently solves problems in physics, is able to synthesize the items on their visual similarity and cope with moral dilemmas.
SME can propose the following situation: because of the dam on the river twenty species disappear. You can save them, if you open the dam. But if the dam open, two fish species will disappear.
Solution SME corresponds to how most people would have received: ceteris paribus act so as to minimize losses; inaction is better than action, which directly lead to someone's death. In the case of dam algorithm will suggest not to open it, because then the two types of fish die directly because of his actions.
A study by researchers at Northwestern University in Illinois (USA) published in the journal Cognition Science.
http://onlinelibrary.wiley.com/doi/10.1111/cogs.12377/abstract
In the article they talk about the basic principles of SME, as well as modern additions thereto that reach a wider range of tasks.
I upvoted You
nice article:)