Word Count: 1,577 Estimated Read Time: 6 Min.

Artificial Intelligence and Natural Language Processing (NLP)

There is so much buzz about AI that most everyone has heard warnings of how artificial intelligence is going to take over the world.  The concern comes with the idea that, thanks to AI, machines can now “learn.”  Some think that if machines can learn, eventually they will become sentient… meaning they will be conscious, aware, and be able to feel.  While some of that is up for debate, the part about machines being able to learn is true and real… now.   With artificial intelligence, there are a wide range of tasks that can be done better and/or faster.  Here is some of what AI can do.

• Computer vision – As discussed last week, AI recognizes and understands objects and scenes in images and videos. This is already being used in applications such as facial recognition, image classification, vision that guides autonomous vehicles and a multitude of other ways.
• Natural language processing (NLP) – AI understands and generates human language, including text, code, and speech. This is now being used in applications such as machine translation, chatbots, and virtual assistants.  It is also being used to write stories, music and more.
• Robotics – AI is being used to control robots and other machines to perform tasks autonomously. This is used in applications such as manufacturing, logistics, and healthcare.
• Deep Learning – AI also gathers and identifies complex patterns from data using artificial neural networks. This is applied in conjunction with image recognition, natural language processing, and machine translation to solve problems, spot trends and identify relationships in data sets. 
• Machine learning – And then there is the skill that scares many.  AI can ‘learn’ from data without being explicitly programmed. This is already being used in applications such as fraud detection, medical diagnosis, and product recommendations. 

Today, let’s look at how artificial intelligence understands and generates human language, including text, code, and speech… and how natural language processing (NLP) is being applied to business.

Natural Language Processing (NLP) Talks the Talk

Natural language processing (NLP) deals with the interaction between computers and human language.  It is one powerful ability of artificial intelligence. NLP algorithms can be used to understand, interpret, and generate human language, which is being used in a wide range of AI applications, such as:

• Customer Service – NLP is being used to develop chatbots and virtual assistants that answer customer questions and provide support.  For example, Norwegian Cruise Line is using chatbots to answer a myriad of questions posed by customers between the time they booked their cruise and the actual departure date, such as questions about passports, visas, connections, ship amenities, day trips, medical support, trip insurance, etc.  The chatbot is able to answer questions immediately at any time of the day or night without having to wait for human help.  This has helped reduce costs for NCL while also improving the customer service experience.

• Product Development – NLP is being used to “listen” and analyze customer reviews and other feedback to identify products that should be improved or identify new potential products. 
  
  For example, one multi-national company that used NLP to listen to its customers was Procter & Gamble (P&G). P&G is a huge consumer goods company that sells a wide range of products, including Tide laundry detergent.  In 2016, P&G began to receive negative customer feedback about Tide. Customers were complaining that the detergent was not cleaning their clothes as well as it used to and was leaving a residue on fabrics washed.  Based on this feedback, P&G acknowledged that there was a problem with its Tide formula and had the company’s scientists develop a Tide formula that was more effective at cleaning clothes and did not leave a residue.  The new Tide formula was launched in 2018.  It was well-received by customers and sales of Tide rebounded in 2019, becoming the best-selling laundry detergent in the US.  P&G’s sales rose in 2019 to $67.7 billion, up from $66.8 billion in 2018.  That is nearly $1 Billion in revenue from a single product.  Excluding the impacts of foreign exchange, acquisitions and divestitures, P&G reported that organic sales increased 5% that year, driven by a 2% increase in organic volume.  P&G’s brand recognition also increased, with a brand value in 2019 of $114.3 Billion. 
  
  Using AI to listen and respond to customer product concerns had a substantial, measurable and direct impact on its sales YOY.   They used computers that talked the talk to ensure that their brand walked the walk.  This shows how using AI to listen to customer feedback in order to improve products can be a very successful business strategy.

• Risk Management – NLP is being used to analyze financial data, legal documents, and other data to identify potential risks and find mistakes. This information can be used to help businesses to make better decisions and to reduce their risk exposure.
  
  Case in point.  JP Morgan Chase has been using NLP to analyze financial data and legal documents for several years. As one of the largest banks in the US, JP Morgan Chase has a lot of data to process.  In 2016, they developed a new NLP system called COiN (Contract Intelligence). COiN is used to analyze legal documents, such as loan agreements and credit card contracts. COiN can identify key information in these documents, such as the terms and conditions of the agreements, the rights and obligations of the parties involved, and the potential risks.  They use COiN to analyze all its new legal documents before they are signed. This helps the bank to identify and mitigate potential risk. For example, COiN can identify clauses in loan agreements that could be interpreted in different ways, or clauses that could expose the bank to liability.  JP Morgan Chase said that COiN has helped it to identify and mitigate billions of dollars in potential risks.  In addition to helping reduce its risk exposure, COiN has also helped JP Morgan Chase to make better decisions. For example, COiN is used to identify loan agreements that are more likely to default. This information is used to make better lending decisions.

• Marketing – NLP can be used to analyze customer feedback, social media posts, and other data to identify trends and opportunities. This information can be used to develop more effective marketing and sales campaigns. 
  
  For example, HubSpot is a marketing software company (SAAS) that helps businesses to attract, engage, and convert customers.  They have been using NLP in a number of ways to improve their own marketing and that of their customers.  They use NLP to:
  
  • Personalize marketing messages – They analyze customer data and identify individual customer needs and preferences. This information is used to personalize marketing messages, such as email campaigns and landing pages.  It is well established that personalized marketing messages are more likely to be opened, clicked on, and converted than generic messages.
  • Target the right audience – They use NLP to analyze social media posts, customer reviews, and other data to identify the types of content that are most likely to resonate with its target audience. They use this information to create and target marketing campaigns more effectively.
  • Create engaging content – They use NLP to generate blog posts, social media posts, and other content that is relevant and interesting to its target audience. This helps to attract new visitors to the company’s website and to keep them engaged.
  • Improve customer service – They use NLP to analyze customer support tickets and other data to identify common customer problems and questions. This information is then used to create knowledge bases and other resources that help customers to find the information they need quickly and easily.

NLP is being used not just in banking, product manufacturing and SAAS.  It is being used in a variety of other diverse businesses and industries including:

• Healthcare – to analyze medical records, patient surveys, and other data to identify trends and patterns. This information is then used to improve patient care, develop new treatments, and reduce costs.
• E-commerce – to power search engines, product recommendation systems, and chatbots on e-commerce websites. This helps customers to find the products they are looking for more easily and to make more informed purchase decisions.
• Financial services – to analyze financial data, identify fraud, and develop personalized financial advice. This helps financial institutions to improve their risk management and to provide better service to customers.
• Media and Entertainment – to generate personalized news feeds, recommend movies and TV shows, and create interactive content. This helps them know and engage their audiences and thus generate more revenue.

Every business owner, leader and manager should be considering how Natural Language Processing (NLP) might be able to help the business be more effective?  If the business is compiling data from customers, interacting with a large volume of customers on routine things and/or dealing with contracts that have language that involves risk, such as insurance companies, then NLP is likely able to help process the data, communicate more efficiently and minimize risk.   It may be time to look at how NLP might be able to streamline processes and cut costs for your company. 

Next week, we’ll examine other aspects of AI that can benefit your business now.  Stay tuned!

Quote of the Week

“AI is the new space race. It’s the next big thing.”
Bill Gates, co-founder of Microsoft

© 2023, Keren Peters-Atkinson. All rights reserved.