In the Pixar movie Up, a fun cartoon dog called Dug wears a magical collar which can detect and translate his barks and cries into fluent human speech. Humans have always been fascinated by the potential to communicate with the animals. This week, an article in the New York Times documented major efforts from a group of researchers using machine-learning algorithms (算法) to analyze the different calls of whales, chickens, bats, cats, and more.

There are several ways to train AI systems now. Typically, Al systems learn through training with labeled data of human language which can be well supplied by the Internet. But analyzing animal language is different. Scientists have to instruct software programs on what to look for, and how to organize the data. This process requires matching gained vocal (发声的) recordings with the visual social behaviors of animals. A group studying Egyptian fruit bats, for example, also used video cameras to record the bats themselves to provide context for the calls.

Many critics of this approach point out two weaknesses of current AI language models: being unable to truly understand the relationships between words and the objects in the real world, and scientists’ little understanding of animal societies. Al language models for humans rely on a computer mapping out the relationship between words and the contexts they could appear in. But these models have their own weak points, and can sometimes be a black box—researchers know what goes in and comes out, but don’t quite understand how the algorithm is arriving at the conclusion.

Another factor that researchers should take into account is that animal communications might not work at all like human communications. There might be unique elements to animal language due to physiological and behavioral differences.

Making a Translator for animals has been a popular project that’s been in the works for the last decade. Although some software has shown some success in identifying the basic vocabulary of certain animals, it’s still a far cry from understanding the complex animal languages.

【小题1】Why do researchers use Al to analyze animals’ calls?

A．To tell the differences among animals.

B．To test Al’s ability of translating animal language.

C．To understand animal language better.

D．To explore the fun of communicating with animals.

【小题2】What makes analyzing animal language different?

A．The lack of labeled data for training Al systems.

B．The difficulty in relating human speech to real objects.

C．The need for sound recordings to provide context.

D．The matching of vocal recordings with their calls.

【小题3】What does the underlined term “black box” refer to in Paragraph 3?

A．Al language models to study animal communication.

B．The researchers’ study on animal societies.

C．The relationship between words and context.

D．The method of Al algorithms to draw conclusions.

【小题4】What is the text mainly about?

A．Al systems for animal language translation.

B．Limitations of current Al language models.

C．Unique aspects of detecting animal language.

D．Challenges in creating a translator for animals.