Led by Ken Forbus, scientists from Northwestern University, Illinois have developed a model that could enable computers to reason and learn like humans, and even make moral decisions. The model involves the use of a structure-mapping engine (SME), based on psychologist Dedre Gentner’s structure-mapping theory of analogy and similarity, which is an idea that an analogy is a mapping of knowledge from one domain to another.
Previous models of SME have been developed, however they were not able to process analogies as complex as those used by humans. This new SME can handle the size and complexity of relational representations that are needed for visual reasoning, solving textbook problems and solving moral dilemmas, according to the university.
“Given a new situation, the machine will try to retrieve one of its prior stories, looking for analogous sacred values, and decide accordingly,” said Forbus.
To encourage more research on analogy and computers, Forbus’ team is releasing the SME source code and a 5,000-example corpus, including comparisons drawn from visual problem-solving, textbook problem-solving and moral decision-making.
This new SME could be valuable across a wide range of applications, including security, health care and education.
Image source: Northwestern University