My best guess: you are over-rating what AI is capable of. Current AI implementations require some "teaching phase" where positive (and negative) "answers" are "learned". Which in effect means that an AI engine is able to reproduce what is has been taught. But this engine is not "intelligent" in the human sense - somehow capable of "self-inspired learning. At least not at the moment.
So, if you are able to understand some animal: teach this to an AI engine. If not: ...
AI needs a supervisor to train it. If some one can fed the characteristics or behaviors of animal into the AI system. Then, it may communicate with animals.
I have given some post regarding the initiation of this idea,
The AI is just a paradigm of calculation. The universe is in the being of the human, if it were outside it will never be reached. Human has imagination, freedom with its limitations: the human being is a creature, his system of integrated comprehension is given ... so he can think only through this system ... the AI can not never reach this possibility ... except with the higher living systems
Humans used to think that they are the top of the possible intelligence. But humans represent a level of intelligence, which is limited. By my opinion the AI will have the possibility for evolution and will become more intelligent, than a human. In some of the areas you can see this high capability. Enter IBM, Watson and Jeopardy keywords into YouTube.
Helping People With Disabilities: Artificial intelligence has also assisted people living independently with disabilities. Voice-assisted AI is one of the major breakthroughs, particularly for those who are visually impaired. It helps them communicate with others using smart devices and describe their surroundings.