Growing up I used to love watching “The Jetsons” on Cartoon Network. Funnily enough, “The Jetsons” used to come on right after “The Flintstones” and we would experience quite the time leap between the two shows.
My favorite parts of the entire cartoon were the flying cars and Rosey the robot maid. Even though she was an old demonstrator model hired by the Jetson family from U-Rent a Maid, I still loved her. The concept of having a robot that understood human speech and could listen to and then carry out commands really excited me.
Fast forward quite a few years and I am living in the age of AI and all the things I used to dream of as a child are slowly coming true all around me.
We are already experiencing the first self-driving cars and I don’t think the ‘flying’ bit will take too long to arrive. Hopefully within this decade. We are now surrounded by multiple Siris and Alexas. All we need to say is “Alexa, play The Good Life” and the song comes on. The hardware doesn’t look like Rosie but we can definitely get some tasks done pretty easily just through voice commands. Now I’m waiting for the implementation of some mechanical capabilities like cleaning and cooking!
Some of the advancements in the field of machine learning and AI in general have been made possible because of big data, which is getting bigger since it is now much easier to capture and share information. A modern car has multiple sensors that share data, and your phone apps have application logs and take pictures and videos too.
All of this data is being produced at such a rapid rate that it isn’t possible for human beings to make sense of it. Thankfully with the advent of big data we also have access to advanced and powerful computing and enhanced algorithms. So we are able to begin to tackle the problem of understanding all this data.
In the case of human to machine interaction and understanding of commands, we use natural language processing.
Natural language processing is a branch of AI that helps machines make sense of human language. This is done through various algorithms based on statistical and machine learning techniques. As with most AI fields, a wide variety of professionals from varied disciplines like computer science and computational linguistics come together to take this science further.
NLP is not a new field but the technology is evolving rapidly because of the availability of large amounts of data. Machine learning algorithms are only as good as the quality and quantity of data that we feed. This increase in text data input has a direct impact on the quality of the output from the algorithm.
But as with all cutting edge ideas NLP comes with quite a plethora of challenges as well. One of the biggest challenges is that text is unstructured data.
Unstructured data is data that doesn’t follow a data model or is not organized in a pre-determined way. Excel files are a perfect example of structured data. Everything is organized in a tabular format, labelled, and easily fed into algorithms and models. In contrast most of the data found in the world around us is unstructured and much of this unstructured data is text-heavy.
It could be the text from a newspaper article, or audio books, or transcripts from TV channels – none of this has an inherent structure to it. In addition, with language and text data it is much harder to create algorithms that will ‘understand’. But slowly and steadily progress is being made and with the advent of GPT-31 hopes are high that we are finally coming closer to our sci-fi vision of human – machine communication.
My dream of having a ‘Rosey’ of my own is looking more achievable with each passing day!
With NLP the goal is to take raw language input and use linguistics and machine learning algorithms to turn it into a form that offers greater value. Some of these value adding applications include:
• Content categorization.
• Topic discovery and modeling.
• Sentiment analysis.
• Speech-to-text and text-to-speech conversion.
• Creating relevant summaries of documents.
• Machine translation of text or speech from one language to another
