Reading the expert's mind, Developing XAI
Indeed, the 14th International Conference on Naturalistic Decision Making (NDM-14) was truly meaningful. Besides meeting NDM researchers coming from various places in the world, I could learn the latest findings and trends in this field.
One of the most thought-provoking presentations at the NDM-14 was artificial intelligence (AI) symposium, especially how NDM studies can contribute to AI's black box issues. It's no doubt AI is a powerful tool in terms of memory storage, calculation, and algorithmic thinking. However, there are some challenges in AI research.
One of them is 'Black Box'. It means both the AI developer and user cannot know the thought process of AI -how AI draws a conclusion. Therefore, human cannot predict how AI will make judgments and decisions. Recent trends in AI research is Explainable AI, also called XAI.
In the perspective of NDM, the main task of our research is to read the expert's mind -how the expert makes a right judgment and decision in the naturalistic setting. The word, 'Naturalistic', means 'real' or 'not artificial'. For NDM researchers, the expert's mind looks like AI's black box. Human thought process is usually invisible.
However, we NDM researchers have revealed the thought process, especially mechanism of intuition, by conducting cognitive interviews. For instance, I have my research partnership with a Japanese biggest acupuncture organization, Hokushinkai.
What we have learned from our research is complexity of human thought processes. There are uncountable influential factors and conditions for judgment and decision. The expert has outstanding information-processing system in mind and can find out a right solution for problem solving.
Intuitive thinking adopted by a human expert is different from algorithmic thinking such as machine learning and deep learning adopted by AI. Nevertheless, there is much similarities between reading the expert's mind and developing XAI in point of making the thought process visible.
I hope NDM research will form the foundation of the future XAI research.