There is a common belief that AI is going to become far more important than it already is, and the truth is that it probably won’t. There is only so much that AI can do before humans need to take over again. Thinking AI is going to take over every function is similar to what people in the 50s thought about how robots would exist in the year 2000. People thought robots would be pushing our carts around the supermarkets when most robots these days are doing little more than cleaning floors and pools. Chatbots, however, are becoming more advanced and is it something we should be worried about?
Are Chatbots Really Learning?
Chatbots are becoming better and more intelligent because people are using them in a better and more intelligent way. The caveman who used a knife to cut up skins for clothing is using a knife more intelligently than the caveman who sticks it up his nose to see if it fits.
It is the same with most forms of technology. Just look at how bad Blender.org was when it was still being used by Pixar. Their old version of that software looks quaint by today’s standards, and yet after years of updates, mods, libraries, frameworks, and much more, Blender is a top-of-the line, and free, 3D animator and video editor.
The same process is occurring for chatbots and has been occurring for many years. It started with the elimination of the most common errors; simple things like crashes, freezes, unrecognized replies, and the dreaded mistake of making people think they are waiting to talk to a real person.
Resting on the Shoulders of Giants
You also have to remember that other technology is progressing in tandem with chatbots, and it creates a scenario where chatbots improve almost by default. An easy, but old-fashioned example, is how people used to program chatbots with different languages, but these days your average multi-language tool is as common as rainwater in the Amazon forest.
Back in the old days, interpreting the variety of different ways people type things and misspell things was a very big problem, but these days your chatbot can fix spelling errors on its own and then spawn the correct answer to whatever issue the user has without prompting the user to correct their spelling. It is better known as natural-language understanding or generation. This sort of thing was a pipe dream just 20 years ago when a single spelling error meant the chatbot couldn’t answer a query.
Chatbots in Healthcare and Customer Service
As you are probably aware, chatbots are used in online messaging systems, forums, and their core programming is even used to automate email replies. Chatbots are commonly seen in the customer service industry because there are many predefined parameters that can be followed. People can ask about sizes, prices, and stock amounts, and a clever chatbot can look it up on a database and answer.
A similar thing is happening within the healthcare profession, from people asking about side effects to drugs, to being able to order their next prescription more quickly and without having to call a busy reception area. A chatbot for healthcare may seem like it is using AI in a very advanced and sophisticated way, but this is because people are consuming well-optimized intelligence. The healthcare industry has many questions that have very easily-concludable answers. Chatbots simply find the problem and offer up the most relevant and correct solution.
Is AI Blooming Anywhere in the Chatbot Industry?
One may argue that the chatbots being used by gambling websites are pushing the boundaries a little. They are gathering information on user behavior as well as user questions and queries, and they are forming their results based on the information being gathered. This is closer to the sort of AI machine learning that Google uses.
By gathering information and using what they learn to approach their online users, they are somehow improving their online experience and lowering the number of people they need in their call centers.
Chatbots are a good way for companies to gather information, the only problem is that most seem to be doing a very poor job of it. Again, gambling websites are doing okay, as is healthcare and even some supermarkets are using their chatbot replies to sell. But at the moment, it seems like chatbot AI is either being woefully underused, or it is being used so ineptly that large and mainstream websites and services are learning nothing about their target consumers.
Perhaps chatbots are getting more intelligent, but for the most part, the people using them are not getting more intelligent, and they are certainly not exploiting the majesty of machine learning to its full potential.