Artificial intelligence could also help us “talk” to animals
One of the great achievements of artificial intelligence has been understanding human language; even writing code based on natural language. AI has not only managed to dominate living languages; rather, with the help of this powerful tool, scientists and historianshave been able to find answers to ancient languages that otherwise would have been almost impossible to discover. However, it seems that the mission of artificial intelligence does not end here, since soon it could help us talk to animals too.
Of course, for artificial intelligence to be able to understand a language, it must first be trained. Currently, AIs learn much more about human language using a vast database. This base can come from various sources; being emails and communication between human beings and virtual assistants one of the largest.
In this way, the learning work of artificial intelligence begins when it studies the way in which human texts are structured and how many times one word appears next to another. Thus, the AI is capable of creating what is known as a “pattern of appearances”; a unique signature that defines the words in a multidimensional parameter space.
The researchers estimate that languages - all languages - can best be described as having 600 independent dimensions of relationships, where each word-word relationship can be viewed as a vector in this space.
WIRED It all depends on context These vectors obey some fairly simple rules, as explained by WIRED. In this way, the authors ensure that “any sentence can be described as a set of vectors that, in turn, form a trajectory through the space of words “. The rules followed by the vectors, for their part, are understood in operations such as: king – man + woman = queen.
These relationships between vectors can persist even when, in a language, a word is tied to related concepts. As an example, the authors talk about the almost 100 words that exist in Inuit languages to describe snow; and it is that all are in similar dimensional spaces . Likewise, when someone talks about snow, they will be entering this same linguistic context.
How could Artificial Intelligence interpret animal communication
Whales For its part, it is expected that artificial intelligence can use the same methods used in human language to interpret animal communication. Of course, this step can only be done once the elements used by the different species in their communication methods have been discovered.
In the case of whales, elephants and other highly intelligent mammals; it would be necessary to detect if they use structures similar to that of human languages in their communication. Thus, an artificial intelligence could come to relate dimensional spaces just as it does with human language today.
WIRED tells us that whales have brains three times larger than that of developed human beings. Likewise, they also have larger cortical areas, and a lower -but comparable- number of neurons. African elephants, meanwhile, have three times more neurons than human beings; although distributed in a different way. For this reason, the medium invites us to consider that this type of mammals could use elements in their thinking; communication and learning processes with which we could connect.
There is still a long way to go Artificial Intelligence At the moment, the elements that these animals use for their communication are unknown. Until now, it is speculated that they could range from tones, to phonemes, passing through blocks of repeated sounds. However, although we do not know the answer, there are already people working to find out.
The aforementioned media mentions the Earth Species Project, which seeks to use all the power of human tools and artificial intelligence to discover and understand the elements of animal communication.
There is something deeply comforting about thinking that the linguistic tools of AI could do something so beautiful; that goes beyond completing our emails and putting ads in front of us, to weave all thinking species.
WIRED Of course, the insecurities generated by the large companies that own the most powerful AIs are not unfounded. After all, they have shown us that their intentions are not always as altruistic as they seem. However, examples like this remind us that this type of promising technology can be used for the benefit of all living beings in the world.