Chatbots vs. Bulgarian Spelling
Can the chatbots change the Bulgarian market? Or their integration will destroy C2B communication?
Brief History of Chatbots
The first computer chatbot’s name is ELIZA and it’s created in 1966. It was created to function as a psychotherapist who developed questions and answers, based on the questions and answers of its patients. Fifty-one years later chatbots are mandatory for a company or organization which is trying to attract customer’s attention.
By definition a Chatbot is a computer program which simulates natural conversation using voice, chat or both.
Many of us relate chatbots with Amazon’s Alexa or Google Assistant, which performs simple tasks, like remembering an event and communicate with the user with an almost natural voice. We’ve all heard of Apple’s Sir, Microsoft’s Cortana and soon to be widely used Bixby by Samsung. They can communicate with us through voice commands or chat. They even have a sense of humor! The chatbots give us unforgettable user experience by exploring and learning from our basic web needs. Sounds great, doesn’t it? Facebook also created a chatbot service for its customers – in Messenger. There are even charts in this category.
Where are we heading?
Through solid investment in development chatbots are quickly evolving from performing simple tasks and conversations to completing tasks, which require their adaptation through self-learning. Soon chatbots will have deeper understanding on the tasks we give them and relatively – deeper competence in their implementation. They will soon be learning from the information they receive from other users or networks (let’s not forget the “Tay” case: The Twitter chatbot which was turned into a racist-homophobic creature in less than a day because of the mischievous users who used them).
The Artificial Intelligence (A.I.) in the chatbots develops rapidly and some people expect to fill the communication gaps between the client and the company. Many organizations are trying to exceed in developing A.I. to offer their clients. Many websites worldwide are already using chatbots in order to improve their user experience. Companies like “Howdy!”, “Action A.I.”, “Cyra” etc. offer chatbots on competitive prices. “Salesforce” created Einstein – chatbot whose role is to improve B2B communication in their platform. Is it useful? We’re about to find out in the next months.
The new startup insurer Lemonade recently released an A.I. chatbot called Jim. Through special algorithms Jim can review insurance appeals filed by the website’s clients. Because of his intelligence, Jim learns from previous clients’ appeals, reviewing them faster each time, constantly breaking his own records. His latest achievement is reviewing an appeal in less than 3 seconds.
The evolution of chatbots is not due only to the development of new technologies, but mostly change in general demographics. The Generation Z and the Millennials and their tech-savvyness represent an increasing part of the world’s population. They expect loyalty from companies, not the other way around. They grew up in the social media world and they know that the information is at the top of their fingers, through their smartphones. That made them anxious and impatient.
This change in demographics is the main reason why A.I. is gaining popularity among companies. Chatbots are an off-beat way to reach the new generation of customers. Although chatbots are still limited in their potential to replace human customer care, they are surprisingly flexible and that makes them attractive to the new generation customers, giving them an unforgettable customer experience. Whether it is a consultant in an online-store, or giving a specific health advice due to a patients’ condition, chatbots will certainly affect the C2B experience in a positive way.
Additionally, as long as the “chatbot-sphere” in the World Economy is growing continuously, thanks to various investors, its product won’t stop expanding its potential, functionality and ability to conduct natural conversation. Bulgarian market is keeping up with the latest chatbot tendencies and we can visit more and more websites, using this technology.
Do chatbots understand us?
We want chatbots to understand us. Companies want chatbots to understand us, because that will make them succesfull. But guess what? Chatbots DO NOT understand us. Science has not found the way that to happen. They still can read only zeroes and ones. That’s the reality. For now… But chatbots can’t understand language nuances. They can’t answer contextual questions.
Chatbots recognize the most commonly used words semantically. If a customer ask the chatbot a question, he will recognize only the known words. The unknown ones will remain ignored. That’s why we won’t get to an understanding easy. Especially if the chatbot is newly installed.
LSI (Latent Semantic Indexing) is an algorithm who helps chatbots to understand customer’s intention. But customer’s intention has to be clear. Chatbot’s “Philosophy” is based on an old principle – similar words has similar meaning. This, of course, is not the best way for text recognition, but it is the best the technology can do at this time. There are also dictionary programs or add-ons for chatbots that help them recognize and “understand” as many words as possible.
Generally when we say “chatbot understands us”, he actually understands that the sentence has subject, scalable and a pronoun. If we are lucky and he recognizes all the words in the sentence, we will get a proper answer from him.
We say that a chatbot is learning if he recognizes the contextual model of a question which is new for him or if he relates it to an “old” question that already exists in his database.
When it comes to a new “context model” chatbots can’t learn it without human help. Or the bot won’t “understand” the new sentence and save it in his “mind”.
Chatbots vs. Bulgarian Spelling
Is it possible chatbots to exist in Bulgaria without losing their artificial minds? I mean the “alternative” alphabet.
When I say “alternative alphabet” I mean the Bulgarian spelling which includes Latin letters, not Cyrillic ones. What do you think – will the chatbots build a bridge between customers and company executives, or they will crush it? Unintentionally, of course.
In the early 90’s, after the fall of the “Iron curtain” new technologies were entering the now free Bulgarian market. Among them was the pager and the first cell phones. Those devices didn’t have Cyrillic alphabet and we weren’t able to write on our own alphabet. Soon things were different, but the Latin spelling became mark of growing illiteracy in our society.
Chatbots are new to Bulgarian internet users and most of them will never understand that correct spelling is essential in order to communicate with others (people or machines). If the chatbots (clever enough or not) were used in a cosmetics shop which has thousands of unique visitors per day, not every one of them would write in the Cyrillic alphabet. This would became a solid problem for the A.I. and the company-owner.
In our hypothetical online-store (we suggest that it is a very big one, unless it does not need a chatbot) would be visited by thousands of men and women every day. The Latin spelling offers its users almost unlimited creativity in writing Bulgarian words. Therefor there are at least 8 spelling options of each word.
For example, I give you the word “Здравейте“ (for our non-Bulgarian readers – it means “Hello”):
- This is the right spelling – “Здравейте”;
- Continuing with the slightly illiterate version – „Здравей те“;
- „Здравеи те“
- Here, we enter the Latin spelling depths – „Zdraveite”
- “Zdravei te”
- “Zdrawej te”
I’m sure you got the idea. Every Bulgarian sentence consists an average 5 words. If each word can be written in 10 different ways, than you must prepare yourself to enter a simple FAQ in 50 different versions just to serve its main purpose – to make the chatbot answer a simple question.
Maybe programming versions of a question in the chatbot’s database is too time-consuming? If so, doesn’t it make the chatbot ineffective? Do we need chatbot if it doesn’t serve its main purpose – to be time-saving, HR-saving and money-saving technology?
Maybe the A.I. will get over this problem someday. But until then I suggest we don’t expect too much of a machine to decipher alternative spellings. May be we should just try to educate humans better first?