Artificial Intelligence Just Got Smarter- Goes Bilingual

0

Neural networks have given automatic language transition a new face. It is an algorithm that is inspired by the human brain. They used huge labeled datasets to be trained to do a variety of complex tasks like the human brain that performs only after gaining the knowledge about something.

Now suppose, you were given Chinese and Arabic books and need to learn to translate Chinese into Arabic. Will you be able to do that? No one among us can. But now, the computer can do that.

                                       Artificial Intelligence

The Breakthrough Story

Yes, you heard it right. Two teams of computer scientists have gained success in translating languages with the use of Artificial Intelligence. Both groups, one from Facebook and other from the University of the Basque Country (UPV), independently created techniques through which neural networks can translate languages. Artificial intelligence can do that without using human intervention or a dictionary. It is all about unsupervised machine learning.

However, its bilingual evaluation understudy score came to 15 in both directions. It is lower than that of Google translate which is 40 or Human who can score 50. But, it is way better than word-to-word translation. Isn’t it? And this is just the beginning. No one knows where this road is headed and what surprises it will bring.

The secret is unveiled

When you hear such wonderful and magical things, a question often haunts your brain- how is it done?

The techniques of both the groups first recognize the pattern in each language. They identify commonly paired words like shoe-socks, tree-leaves, table-chair, etc. that are common across the languages.

Once these patterns are recognized, the neural network then links these co-occurrences in both the languages. This develops a bilingual dictionary on the accuracy of the translation, without any human feedback. Then, these dictionaries are used for translating the whole sentences.

It is more like a giant atlas with words for cities. The maps in each language will resemble each other with just different names. All computers do is figure out a way to overlay one map over another and voila, you have a bilingual dictionary ready.

Apart from this technique, neural network uses two more ways- back translation and denoising.

In back translation, a sentence is translated to the required language and then back again to the original language. If there is any discrepancy between the two sentences, neutral networks adjust itself and then tries to make a more accurate translation.

We Taught AI Racial And Gender Biases

0

Programs like Google Translate has experienced a dramatic hike in the ability of language interpretation in past few years. Thanks to the new machine learning techniques and of course to the numerous online text data. It is on these data that the algorithms can be trained.

Everyday machines are acquiring the human-like abilities of language. Along with it, they are also inheriting the deeply ingrained biases hidden in the patterns of language. AI has been seen exhibiting a striking gender and racial biases. Many studies have proved that AI is biased towards gender and races. Is it really biased?

                                    AI

Is AI Biased Or We Are?

People say that this shows AI is prejudiced and biased. No, it shows we are biased and prejudiced and AI is learning it. Because we are assisting AI in its learning. It is reinforced in AI because algorithms are unequipped in consciously counteracting the learned biases, unlike humans.

Embedded Biases

Machines learn from Word Embedding. This process has transformed the way computers interpret text and speech. This method has successfully helped computers in making sense of the language in past few years.

Word Embedding builds up a mathematical representation of language. The meaning of a word in it is distilled into a series of numbers, called Word Vector and based on which related words and terms words frequently appear together.

For example, in the mathematical language space, words for flowers and pleasantness are clustered closer while words for insects and unpleasantness appear close together. Thus, to AI flower is connected to pleasantness and insect to unpleasantness.

This purely statistical approach has captured the social and cultural context of words differently than a dictionary. Some of these implicit biases in the experiments of human psychology has been captured by these algorithms. Words like female and women became closely associated with arts, humanities, and home while the words male and the man appeared closer to the professions like Math and engineering.