Conversations with your computer will be easier than ever

Source: Google AI Blog

Language is the key to human evolution, Language has helped us gain information, share information, and build on it. Who knows? without the power of words and spaces, the human race would have been extinct even before the dinosaurs, Yes, the end of the human race. OK, Let’s not get that dramatic.

Even with great vocabulary, language fluency, and a beautiful voice we sometimes fail to express ourselves and we end up feeling lonely till we find a book, a movie, a work of art, or a computer that talks to us in our language. Yes, a computer. A computer talking to humans is not fiction and we know that with the advent of Google Assistant, Siri, Alexa, etc. But having lengthy conversations as you would with a friend i̶s̶n̶’̶t̶ ̶ wasn’t a reality until now.

Google’s latest research breakthrough called LaMDA- “Language Model for Dialogue Applications”, will let you have conversations with your computer at length.

#TECHFORALL- A series by The Vinning Story

Why is having conversation with a computer difficult?

Machines don’t talk any of our languages, they talk in bits (1’s and 0's), we understand that and have developed technology for us to talk to machines and also for them to understand us enough to give us results. The technology to make machines understand our language the way we do is a mammoth task, it would take machines to understand the nuances, the context, the metaphors, the ironies, touches of sarcasm, and a lot of proses, idioms, figures of speeches; you feel the fatigue. And we need to make them understand these things in their language of bits, not using a Wren & Martin grammar book. This process in the language of technology is known as Natural Language Processing (NLP), this has been in place for years helping you find answers to your questions on Google, searching YouTube videos, finding your products on e-commerce websites even when your words are misspelled and sentences are unmethodical. But, NLP has limitations that make it hard for Human-Computer Interaction beyond a certain degree. Researchers from around the world have been developing techniques to overcome the challenges of NLP.

How does LaMDA overcome the Limitations of NLP?

Natural Language Processing (NLP) is a branch of Artificial Intelligence that aims to ease Human-Computer Interactions (HCI). NLP works by training a system using an enormous amount of data called the training data. But, the requirement of the enormous amount of training data is a big hurdle, so researchers from around the world have been developing a variety of techniques to make the systems more efficient to be trained with small-data. One such technique called Bidirectional Encoder Representations from Transformers (BERT) was developed and open-sourced by Google in 2018. BERT was trained to recognize, interpret words up to a few paragraphs which aren’t enough for holding conversations with a Human.

LaMDA came into existence to overcome these shortcomings of its predecessors. The model was trained with dialogues, enabling it to have open-ended conversations by understanding the nuances of the language. LaMDA through its training developed abilities to provide sensible, and specific responses during a conversation.

Photo by Waldemar Brandt on Unsplash

Example:

Input (Human): I feel less productive today, I feel guilty of wasting my time

Response (Models trained with words, sentences): Do you want me to suggest a few productivity apps?

Response (Human and LaMDA): Don’t be too hard on yourself, we all have those days. How about working on the task now? (Generic Example)

And the conversation goes on, like a conversation with a friend, like in a screenplay. That is because of LaMDA’s nuanced understanding of language and its ability to respond with Sense and Specificity. You can soon have many meaningful, long conversations with LaMDA that too without being judged.

Just say, Hello Google!

--

--

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store