This paper reports our new idea as a suggestion for improving Artificial Intelligence, especially the field of “Utterance-Unit Tag.” Rapid technological innovation develops the high artificial intelligence in the area of discourse. To improve this area of skill more, such as adopting the “Utterance-Unit Tag” to the description of our actual, subtle utterances, AI researchers definitely need to classify the concept of “illocutionary forces,” proposed by J. L. Austin （1911-60）, more closely. Now the model based on J. Searle （1932-） ’s interpretation of J. L. Austin’s “illocutionary forces” is the generally accepted approach to developing AI discourse, according to previous research. Although Searle emphasizes our intention and function of the utterances, J. L. Austin’s model, however, is more appropriate because Austin focuses on not only verbal convention but also physical convention in his speech-act theory.