How Can Machine Learning Improve Chatbots?


Machine learning chatbots may seem like a new trend. However, there is a storied history dating back to 1950. In an article, Alan Turning wrote, “I propose to consider [a] question,” he wrote.” ‘Can machines think?'”. He details the Turing Test, a technique to gauge whether or not someone was speaking to a person or a chatbot. In 1966 Joseph Weizenbaum developed Eliza 1966, regarded as the chatbot. Although she blazed the trail for future technology, she failed to pass the Turing test. Another bot, A.L.IC.E. (Artificial Linguistic Internet Computer Entity), created by Dr. Sbaitso, also failed the Turing test in 1995, although still regarded as the most advanced chatbot to that date. A.L.I.C.E. could use proper inflection and grammar. Unfortunately, her voice was far from sounding human.

Watson, designed by I.B.M., defeated several Jeopardy champions in 2006. In 2006 Watson, a breakthrough in Machine learning and neural languages.

Machine learning can improve chatbots by giving them the ability to respond to queries in ways that do not require the designers of the chatbot to what the query will be. Therefore, a chatbot enhanced by machine learning will be able to converse with users more naturally than ones without.

Predominantly, chatbots have a female voice. According to Anne Ahola Ward, writer of The SEO Battlefield,” I’m not sure if it’s 100% true, but I’ve heard from many sources that the main reason most computer voices are female is because of the movie 2001: A Space Odyssey. Allegedly, computer voices were once predominantly male, but the computer voice of H.A.L. freaked people out so much that for the next 20 years, machine voices became increasingly female. Siri, Alexa, and Cortana, and their predecessors show this might be true.

Major Chatbots Currently in Use

Siri launched in 2010. She was the first chatbot to provide not only personal assistance but also a lively tone. In 2011 Apple quelled some of Siri’s questionable traits and added multi-lingual features. Additionally, she became available globally and could give verbal answers instead of written ones. Indeed Siri paved the way for the following chatbot assistants.

The roots for Microsoft Cortana started in 2009, and her official launch was in 2014. She shares her name with a Halo character, the inspiration behind her technology. She initially designed for Windows Phone 8.1. Over time, she became a staple in all Microsoft devices, including XBoxes.

Amazon’s Alexa followed in 2014. The name stands for Artificial Language Experimental Device, initially promoted as a digital music source. Then, she graduated to the echo and became a one-stop to run a household. Currently, there are many different things Alexa controls, including vacuums and vehicles. Besides her functional features, she can sing, play games, give compliments, and many other things. The standout feature of Alexa was her ability to understand customers, until this time a skill lacking in her predecessors.

Microsoft Cortana was the second chatbot launched in 2014; she shares her name with a Halo character, inspiring her technology and Windows Phone 8.1. Over time, she became a staple in all Microsoft devices, including XBoxes.

Additionally, we’ll use voice to text which is now commonplace when we don’t have thumbs to text. Or, if we are visiting a foreign country, Google Translate is a go-to during a language barrier.

In an article written by smallbizgenius, 1.4 billion people are using chatbots. Additionally, they can handle 80% of the questions asked—34% of customers preferred utilizing a chatbot over speaking to a customer service representative.

I.B.M. defines Machine learning as “a branch of artificial intelligence and computer science that focuses on the use of data and algorithms to imitate the way humans learn, gradually improving its accuracy.” UC Berkley broke these algorithms down into three parts. First, the decision process. These are the learning algorithms used to make a prediction or classification. Labeled or unlabeled, they can produce a pattern in the data. Next is the error function. It evaluates the prediction of the model and assesses the accuracy of the model.

The last piece is a Model Optimization Process which fits the training points into the training set then adjusts to reduce discrepancy, repeated until a specified accuracy target. Machine learning is gained by a process called natural language processing or N.L.P. Experts agree this is only a beginning. For chatbots to be fully functional, they must give an informed answer, maintain a conversation, and be unable to be differentiated from humans. Referring back to the Turing test, this last characteristic didn’t pass. Yet, as demonstrated in homes, cars, while on phones, and stuck in traffic, we have become more willing to utilize the technology and practices that chatbots offer.

The reason for machine learning chatbots

Petter Bae Brandtzaeg wrote in Why Do People Use Chatbots?, “Chatbots are important because you won’t feel stupid asking important questions. Sometimes talking to someone can be a bit intimidating. Talking to a chatbot makes that a lot easier!” Although this is not always the case with automated services in customer service. It’s becoming standard to call a customer service center and reach a computerized machine.

As the technology grows, it’s become increasingly easier to ask a simple question and contact the correct department, or in some instances, have your question answered without human interaction. Yet, when it comes to complicated questions that require longer answers and fewer data-based responses, chatbots are still having a tough time keeping up. One of the disconnects is that some companies have failed to consider their customers.

The chatbots have learned to ask again if the request is not understood. They can still not direct the conversation to the solution. They can make the interaction daunting because a customer has to ask a question numerous times to get the correct response. Worse, the case has to demand a customer service representative and is angry at the inadequate response. One of the most cited issues is a disconnect in I.T. teams and failure to update the systems.

Although the technology is still in its fledgling stages, any company that implements these services must be willing to use chatbots organically and use the prompts and information gleaned from the customers to improve the intelligence and functionality.

Since there are so many pitfalls and benefits to chatbots, does machine learning can benefit this technology? According to multiple sources, many things need to be corrected. As mentioned, since these systems are designed using algorithms, there needs to be a more significant focus on how the customers speak to these devices.

There are better success rates in fields like real estate, banking, and the daily task of bill paying is that despite this device still not sounding like a human, they can easily recognize commands that call for logic or fact-based answers.

However, when companies that focus on complex customer issues try to utilize this same technology, there are problems. Currently, chatbot technology learning rates are at a middle school level and need far more programming to reach full functionality. This is a steady progression with businesses small and large using this service. It allows stores and online businesses to have 24-hour service and cuts the cost of running a business. Consumers can order a shirt, a pizza or make sure the lights are on when they get home.

And still, sometimes these services have to ask the same question repeatedly, number spoken are not recognized, and customer service representatives become the defaults.

Machine learning is improving functionality and creating better algorithms for ease of usability. The iconic trio of chatbot assistances has been mentioned along with the benefits of businesses. The gap of repeated “I” m sorry I didn’t understand,” or wrong information recited being repeated back is closing. This is undoubtedly due to machine learning.

The Busness Value of Chatbots

Businesses are turning towards chatbots to assist their clients.

Real Estate, finance, healthcare, travel industries are benefiting the most. 80% of all businesses will utilize some form of chatbot by 2021 (outgrow 2018). There is a multi-generational trend that prefers chatbots to human interaction. However, many are still highly put off by the robotic, barring on mocking voices.

Machine learning technology uses an approach that helps it learn better than previous technologies based on better algorithms. Yet, it’s a challenge to keep these systems moving along with them with the fast-moving tech world. Numerous passes need to be made, and as this technology spreads, more and more data is required.

The personal sector of machine learning seems to have better reception than using it while trying to elicit information to solve things like tech support issues or other questions with more complex answers.

Peter Gentsch stated, “To the user, chatbots seem to be “intelligent” due to their informative skills. However, chatbots are only as intelligent as the underlying database.” Therein lies the Cruz of the issues. Building and evolving the databases and algorithms to accommodate reliance on chatbots sometimes adds more complications than benefits. There are so many components and considerations that need to be addressed. Financial institutions, retail stores, and other businesses trying to provide better support to their customers have certainly benefited from this technology. Unfortunately, machine learning cannot fully cover nuanced questions.

As this technology evolves and grows, it will surely make huge advancements. Undoubtedly, one day it will be indistinguishable from a human voice. The pioneers’ visions of capabilities that far exceed what we would believe possible are on the horizon.

Machine Learning & Chatbots

So, does machine learning improve chatbots? Absolutely. This technology creates a neutral dialogue between people and A.I. It has become a relaxed way for people to unwind, get there new, ask for the weather, and a myriad of other tasks without leaving the comfort of their couch. This technology also eliminates the last-minute panic of a business closed and a bill that is due. Alongside the comfort and ease is the ability not to overthink a question or wonder how to find the answer. The evolution of machine learning streamlines tasks far more efficiently than the proceeding technology.

Technology that supersedes machine learning is already being developed, and perhaps chatbots will go the way of something fresher. Yet, this is a product of our technologically driven society. The world is fast-paced, and it’s only natural to constantly search for the best way to make our busy lives more efficient. Some might not remember when a chat room was the most revolutionary entertainment or AOL disks populated the mail like the owls that brought Harry Potter his Hogwarts letters. In the tech-driven world, this is to be expected.

Devices and services powered by machine learning chatbots are part of normal daily functions and serve so many purposes. Utilizing this technology means we no longer have to come home to a cold, dark house. We can monitor our residences from three thousand miles away. There is no longer a frantic run to check the inbound and outbound flights. Countless other examples can be cited.

Being able to interact with technology of this magnitude was the stuff of dystopian fiction. It may have once seemed impossible to have a device understand and interact with us much the same way a friend would. The algorithms and constant evolution of machine learning make these things possible. It’s also fascinating to think how we are further advancing machine learning with each “Hey Siri” or “What’s the Weather Alexa.”

Chatbots in Our Future

On their website, I.B.M. stated, “Over the next couple of decades, the technological developments around storage and processing power will enable some innovative products that we know and love today, such as Netflix’s recommendation engine or self-driving cars.” When Back to the Future made its theatrical debut in 1985, things like flying cars seemed lifetimes away.

Gen Zers may not remember a world that predates automation and fast internet speeds. Machine learning chatbots were built on technology found in pioneers like Sophia, the first humanoid robot to gain citizenship and the only nonhumanoid to attain the United Nation’s Development Program first-ever Innovation Champion. Now that we have become a global community, platforms like Facebook, Snapchat, Facetime are the preferred methods for social interaction. Each time we use the chatbot features, we contribute to machine learning fueling the growth of chatbots.

Rashid Khan wrote in Build Better Chatbots, “You all know that chatbots are a new technology altogether. It’s like the early age of the Web. Things are still shaky yet growing at the speed of light.” The advancements in the last half-century have shown that machine learning will fuel chatbots and productivity with consistent work and collaborative effort. It is equal parts technology and recognizing the need to use each interaction to evolve the platforms. It’s not just large corporations taking advantage of this advanced technology.

Small businesses are seeing the necessity and how this can increase productivity and make their customer interactions more meaningful. There are industries where this type of technology is not beneficial, but they are outnumbered by those who have found it an integral part of their daily operations. The analytics on our phones, how our Alexas learn our voices, being able to ask for and receive directions without numerous queries are all parts of machine learning.

Each step and interaction creates the data, algorithms, and knowledge for machine learning to evolve these chatbots to make them productive additions to our ever-evolving world. Machine learning must become a collaborative cross-platform technology to thrive. And, with the shift towards this in everyday events and daily necessities, this is a reality.

Conclusion

The trends predict that machine learning will continue to draw and create databases from information that predates the technology we rely on today in the following years. It has already had a rising trend with smartwatches and other wearable devices. Therapy apps that help you stay mindful are another excellent use of this technology. During COVID-19, the true worth of this technology was put to the test. People confined to their homes relied on Facebook Messenger, Snap Chat, and other chatbots to stay in communication.

It’s safe to say we interact more with the devices that assist our daily lives than utilizing people who perform similar functions. Some might say that machine learning has created a disconnect in human interaction. That is only partially true. This technology allows us extra time to spend with our family, even if it’s on FaceTime calls. Machine learning streamlines grocery order, keeps us updated on appointments, and adds a smoothness to our lives that were not available before this technology became a regular part of our lives.

Gene Botkin

Gene is a graduate student in cybersecurity and AI at the Missouri University of Science and Technology. Ongoing philosophy and theology student.

Recent Posts