Artificial intelligence (AI) is changing the functioning of various industries through smart and quick characteristics such as the ability to identify any object in mere seconds, imitate the human voice, carry out administrative work, give suggestions and much more. But often the intelligent machine lacks the most basic knowledge of objects and situations. If AI wants to execute a task perfectly, it needs a brain of its own. To grow this brain DARPA is teaming up with the Seattle-based Allen Institute for Artificial Intelligence to carry out the initiative “The Machine Common sense program” which will test and teach AI common sense.
This is more complicated than adding a few programs into a machine, it will require sophisticated constructs involving multiple concepts and intuitive connections. This is why DARPA’s concept includes creating “computational models which will learn from experience and mimic core cognition domains as described by the psychology of creation to test and teach AI common sense. It includes the domains of objects (intuitive physics), places (spatial navigation), and agents (intentional actors).
This approach will first test the ability of the machine by presenting it with written descriptions of situations and several short options for what happens next. 1,13,000 such questions will detect how much the machine knows. Gradually a possibility to overcome challenges such as recognizing a photo or face of a person and identify the expressions will be made. The common-sense initiative is part of DARPA’s massive investment of $2 billion in AI on several fronts. But they don’t aim to imitate or compete with the likes of Google, Amazon, and Baidu who have invested heavily in the small AI applications we see on our phones and the like.