How can NDM contribute to AI research & development?
Unfortunately, it's hard to say that research on Naturalistic Decision Making (NDM) has pervaded across Japan and the rest of Asia. Last year, I attended the 14th International NDM Conference (so called, the NDM-14) held in San Francisco. However, I have never met any researchers from Asian region (except one US graduate student who was originally from India).
One of the most attractive aspects of NDM research is that NDM is applicable to any field -military, aviation, emergency medicine, petrochemical engineering, sports, business management, legal matters, fire fighting, police, and much more - as far as human experts make their decisions on the spot. We NDM researchers especially focus on how the experts make a judgment and decision under stressful and imminent situation.
At the NDM-14, I met university professors and researchers as well as some private consultants who specialize in NDM and macrocognition. The consultants undertake the inquiries from the clients about developing training programs for new and intermediate-level staff members.
For instance, the US Defense Advanced Research Agency (DARPA) has its project to develop the explainable artificial intelligence (XAI). All Japanese people and other Asian don't know the fact that Gary Klein and other NDM researchers join this project. Our findings leads to the direction of XAI research.
XAI is not a human-like AI robot that can explain about its thought and make an independent decision. It is a decision aide for human experts and other ordinary people. It's true some types of AI surpass human experts in certain fields. Deep Blue and Alpha Go beat the world's chess and go masters. However, it does not mean that AI is omnipotent and will replace human experts in all fields. In fact, professional chess and go players still exist.
Through our research, I can confidently state the following four points:
1.) Human cognition (i.e., perception, awareness, sensemaking, reasoning, analysis and logic, calculation, memory, intuition, insight, pattern recognition, simulation and imaging, decision making, explaining and communication...) is much more complicated than we think. We cognitive psychologists do not understand each of cognitive functions exactly.
2.) AI is superior to human experts to calculation, memory and algorithmic thinking only. AI could outperform human experts as far as accurate calculation, strong memory and algorithmic thinking are necessary for the tasks (such as chess and go).
3.) AI researchers have not developed new types of AI that contain other human-like cognitive functions other than calculation, memory and algorithmic thinking yet. Many of our jobs and even trivial tasks in our daily lives require various cognitive skills besides these three areas.
4.) For human-centered AI development (NOT machine-centered development), understanding human cognitive system is prerequisite. Both humans and machines have strengths and weaknesses. Human-centered AI supports human experts for their better performance -more accurate judgment and enriching creativity, not replacing humans.
My serious concern about AI research in Japan is that the direction of research is machine-centered development. Last year, I attended several big AI conferences. However, any AI researchers did not refer to NDM and macrocognition. Rather, their research interest is how to develop superhuman-like AI robot.
My duty as only the Japanese NDM researcher is to show the correct direction of AI research by introducing NDM / macrocognition to Japan.