Summary – This article explores the origins of the chatbot, the predecessor to today’s ChatGPT and other forms of generative AI.
Over the past few months, talk of ChatGPT and Google Bard have reached a crescendo. As this type of AI ventures out into the mainstream, people who may have never uttered the word “AI,” now speak of it casually. But while the future of generative AI is being framed and its role in how we work and live is being sorted out, there’s a form of the technology that has been around for years, which continues to play a key role in businesses everywhere – it’s the old fashioned chatbot.
Chatbots are programmed to simulate a conversation with a person – either by voice or text, using natural language processing (NLP). At its most basic form, a chatbot answers questions based on rules on which it is trained. In a more advanced form of a chatbot, it is trained on key datasets and rules, yet because of its AI capabilities, it becomes smarter based on the interactions it has with the end users and can answer increasingly more questions or provide the same information that a human can – if not more.
The first chatbot was developed almost 60 years ago at the Massachusetts Institute of Technology (MIT)) by Joseph Weizenbaum. He called it ELIZA, after the character from the George Bernard Shaw play, Pygmalion. ELIZA would rephrase whatever speech input it was given in the form of a question. That was about the level of understanding the first chatbot had, yet people were impressed with a machine that could communicate with people, and that interest has only grown since then.
Today, the goal of all chatbots is to accurately imitate human interactions, logic and intelligence, yet while they still have a long way to go to match human characteristics, there are key technologies that are enabling their continued evolution.
The rise of the cloud: Because of its adaptability and availability, companies have been able to more rapidly create chatbots that can integrate with other systems, such as customer relationship management systems (CRM), enterprise resource planning (ERP) systems and others, to access information enterprise-wide and accurately answer questions.
Natural Language Processing (NLP): Built upon machine learning, this is the driver to conversational AI and helps chatbots understand the nuances of different accents, phrases, dialects, and even languages, so that they can communicate and interact with humans.
Artificial Intelligence: While the earliest forms of chatbots were not built upon AI, today AI is empowering smarter chatbots that are able to detect changes in tone and human sentiment and respond accordingly. It also enables chatbots to predict what may happen next based on historical data to make more data-driven choices.
Consider some of the ways that chatbots have been empowering businesses to operate smarter:
Call-center support: Chatbots have been widely used by customer service departments, addressing customers’ standard questions about their insurance policies, government licenses, software problems, health issues or store returns, to name a few. By providing support 24/7 companies are able to provide service more consistently than they could with humans only, and allowing humans to focus more on those questions that the chatbots escalate up the chain, since they may surpass the chatbot’s capabilities.
Retail engagement: In retail, chatbots are optimizing brand engagement and sales and keeping customers on a particular retailer’s site longer. While chatbots that randomly pop up and ask questions to engage with you can be pesky at times, customer feedback is allowing retailers to fine-turn their use so that they are less intrusive and provide valuable information that consumers care about.
Higher Education: Chatbots have been used quite successfully to address student questions via social channels and SMS mobile texts. For example, the Inter-American University of Puerto Rico deployed a chatbot that was able to automate 80 percent of the questions it received daily from students, while cutting in half the number of support staff required. This enabled staff to focus more on the tougher questions and proactively work with students.
While rule-based or NLP-based chatbots have been playing a key role in different industries for many years now, the next step in their evolution is generative AI, in the form of Chat GPT, Google Bard, and the list of players will continue to get longer. With generative AI, developers train chatbots on massive data sets, which help them understand natural language better than ever possible with its predecessors. They can scour the internet to generate creative content, develop software code or develop email campaigns within seconds, at the prompt of a user.
Despite all the clamor, generative AI is nothing new, it’s just the next step in the chatbot journey. And, as with its predecessors, today’s generative AI needs to be approached carefully and with some level of trepidation. It was said that even Weizenbaum, the founder of chatbots, was leery of giving machines too much responsibility. AI technology is going where it never went before, and has the power to do amazing things, but also to cause harm if left unchecked. Perhaps Weizenbaum and his chatbot was on to something.
How do you view this new generation of chatbots and their role in the world? We’d love to hear your thoughts. Email us at email@example.com.