I think it is just one of the option for data analysis. It is still popular tool for those who have no sufficient abstract mathematical background. It is still mathematics but low level and easy for basic mathematical background researchers.
I feel that it might be convenient in some cases but it certainly isn't necessary because there will be always an injective mapping from any set of multivalued states to a set of binary coded states.
Fuzzy Logic generates continuous-data (real numbers) outputs; while Boolean Logic generates discrete-data (binary) outputs. In Artificial Intelligence (AI), both of these outputs (continuous data and discrete data) are very relevant and vital to the decisions and/or predictions proffered by AI-based systems.
In fact, many AI-based systems rely on either continuous-data inputs or discrete-data inputs, or both. In other words, many AI-based systems depend on either Fuzzy Logic or Boolean Logic (or both Fuzzy and Boolean Logics) for inputs/outputs.
Therefore, Fuzzy Logic is still necessary and relevant to Artificial Intelligence.
Bonaventure Molokwu : But if AI would vitally depend on the use of continuous values, then it could work only in theory because any implementation on digital hardware involves necessarily the discretization of the input and output values, and each discrete number is represented by an (arbitrarily large but finite) set of binary values.
While the errors introduced by discretization can be treated as noise which can be made arbitrarily low by increasing the number of bits used in the representation, the implementation on analog computers would be hampered by (physically generated) noise which cannot be decreased arbitrarily.
Fuzzy logic, sets and systems is not just a simple part of mathematics as some people think. See, for example, the impact factor of the journal Fuzzy Sets and Systems, that is printed by Elsevier.