From Austen to AI
How a love for reading makes better chatbots
Written by Sara Gilsdottir and Helle Lindheim Stavsholt.
When you hear the words artificial intelligence, your thoughts will probably not wander to dusty books, Aristotle or liberal arts degrees. Naturally, as two literature students, we didn’t imagine a couple of years ago that we were suited to work in a tech startup or that a tech startup would be the most sensible employer for us, but here we are, working in a company that develops artificially intelligent chatbots. So, which tasks are our responsibilities and what do we bring to the table?
When we tell people that we work with chatbots, they often take for granted that our technological knowledge is sky high. This is understandable, but our contribution to the company isn’t technological competence. As UX writers, we design the words people see when they interact with our chatbot platform, Kindly. When meeting clients (brands or companies who want to include a chatbot on their web page), it is actually a strength that our references are not tech specific as our customer base is not limited to computer scientists — making the ability to communicate our technology without advanced terminology important. Most of the people using our platform, ourselves included, have never ventured into coding, and programming can easily seem like a wild jungle of 1 and 0s. Hence, we may view and use our platform in other ways than our tech colleagues, which is advantageous. Through collaborating and exchanging points of view, we can implement better solutions and continuously develop our platform.
A question literature students encounter on a regular basis is “what is your future job, then?”
We have to admit that this question is warranted. It’s true that studying literature doesn’t pave the way for a specific career. However, our education makes us flexible, and it’s not necessarily our titles that determine which jobs we get, but our acquired abilities.
Flexibility is key as every chatbot project is different. They each have their own unique content which is to be implemented and structured to a chatbot format, and they have different target audiences they would like to reach. How the company decides to communicate with its customers is up to them, as there is no correct way to address the end users. One golden rule is to create a chatbot identity that is consistent with the company’s other channels. In these kinds of projects, our abilities gained from studying literature comes in handy. Why? Because the professors in our studies often evaluate our work based on how we structure, argue, gather relevant information, think critically and analyze historical epochs or genres. Through reading and discussing literature, we’re constantly developing the ability to see information from different perspectives and become familiar with various ways of expressing ourselves both textually and verbally.
Let’s say we were to create a chatbot answering the question “When was Pride and Prejudice published?”. Even though our machine learning model is clever, it needs good input to function properly. We as UX writers would provide the chatbot with different ways of expressing the question, mimicking how end users would phrase themselves: “was Pride and Prejudice written in the 1800s?”, or “which year did Jane Austen write Pride and Prejudice?” are examples of what an end user might write, and hence what we should write as input to our machine learning model. When reading is your main domain, you will regularly collect new words and phrases to integrate in your own language. A substantial vocabulary is beneficial in order to feed our machine learning model with quality input.
When the chatbot is trained with sufficient data, it needs to provide grammatically and factually correct responses. Several years of writing assignments and examining texts have rendered us sensitive to grammatical errors, which is helpful when writing documents used for internal or external training, social media posts or other types of texts that require a thorough knowledge of grammar and structure. As UX writers, we also structure answers, buttons, and links in order to make the bot understandable and organized, both for the people working on it and for the end users.
Imagine you’re an author who has recently written a novel you’re exceptionally satisfied with. You’re eager to see the manifestation of your hard work — the physical book — in the local bookstore. For this to happen you will need someone to complete tasks such as reading the script and give professional feedback, design the cover, write the blurbs and print the pages. This distributing of roles also applies to tech companies.
In Convertelligence, our developers are the authors. But instead of writing similes and tropes, they write code. They’ve created Kindly, and in order to see the product in the bookstore (or on a web page in our case), Convertelligence needs to inform potential customers that we exist, introduce clients to how it works, provide the bots with precise language and correct grammar, keep track of the economy, create good design solutions and so on.
Educational diversity provides different experiences and mindsets to our company, and is an absolute necessity in order to make the whole machinery function properly.
In the end, technology boils down to human beings, placed at the creating, developing and receiving end of technological inventions. The app you downloaded a couple of weeks ago, the radio in your car and your beloved coffee machine all have something in common: they are creations which goal is to make your everyday life easier by solving your problems or creating efficient solutions. Technology is a tool used by people, not machines. This is what makes knowledge about people, language, communication and an understanding of the world’s history and current state valuable within the tech field.
Convertelligence is now known as Kindly. Visit our website to see what else is new!
