A look at AI today
Not very long ago, in my student days, there was one particular activity each semester that I didn’t really look forward to. The drawn-out student evaluation survey. It would have been very nice to have something more interactive, fun and high tech that maybe makes use of Artificial Intelligence (AI). The other problem was I often felt like no actions were taken in regards to my surveys, or at the very least I didn’t know if anything was done.
With every year, it seems like AI is taking a bigger and more sophisticated role in our lives – including our working lives. It has made its way into the realm of employee feedback, where high-tech, AI-backed solutions are essential to understanding exactly what employees think. These solutions are more efficient than traditional methods such as onerous annual employee engagement surveys and allow for continuous feedback collection and action in real-time. Feeling like I am well listened to will definitely make me more likely to stay at a company.
In addition, these more traditional employee surveying methods undoubtedly mean thousands of sentences of text to analyse, and that doesn’t even include the quantitative scores. The most meaningful employee feedback is qualitative, especially when it comes to feelings expressed about something specific, or suggestions on something which could be done better. Technology that processes these answers efficiently is highly in demand.
Recently, machine learning and natural language processing AI systems have been on the rise. The basic premise behind these algorithms is that they are able to extract the critical data from open-ended questions, identify patterns and use keywords to figure out the sentiment of the respondent. They have big implications because they are able to do this for very large datasets. This is also known as text analytics, or ‘text mining’ and will be discussed further on in more detail.
Chatbots and employee feedback
Meanwhile, a wide variety of companies have utilised chatbots, primarily to make customer service more efficient, and allow for the quick answering of the most important or frequent questions. A chatbot is a computer program that automatically answers questions and ‘makes conversation’. It can take use of several mediums, such as websites, SMS, social media and spoken language. One way of designing a chatbot is to encode as much data as possible into it, so that it effectively cannot be discerned from a person sitting behind a computer. Another is choosing a selected handful of questions that can be answered by the bot. Chatbots generally have a given purpose or topic that they can ‘talk about’ – something to do with the company or person’s operations, therefore the second design described above is usually more practical. For example, Apple has the well known chatbot Siri, who serves as a virtual assistant. I’ve seen and interacted with many chatbots on e-commerce and webstores, but have not seen them too often for other services – though I believe they could be quite helpful.
One such service chatbots could help me (and like-minded millennials) with is employee feedback. As I mentioned before, technology like this would make the whole process of giving feedback much more engaging. For this purpose, chatbots could be used to capture feedback more naturally and conversationally, allowing people to say what’s important in their own words. The Evolved Group, via the PeopleListeningTM platform, has developed a chatbot, named EVE, to be one’s ‘feedback companion’ at work. The idea behind EVE is to combine the functions of a two-way conversational chatbot and a text-analysing AI tool. EVE employs machine learning to recognise what employees want to talk about and then asks relevant questions. As such, allowing employees to give meaningful insights in their own words is key.
While EVE sits on the side of someone’s screen making work-related conversation at timely intervals, the answers are analysed with a process called ‘text mining’. This can mean anything from removing ‘stop-words’ between keywords, ranking and assigning sentiment scores to key topics. It can especially help with segmenting employees by preferences and satisfaction.
Sentiment analysis is important because it helps figure out what’s missing and what matters most to employees. It gives a relatively narrow band of words (compared to whole English language) a significant weighting. For example, words like ‘good’, ‘timely’, ‘effective’ and their opposites are highly weighted. Adjectives like ‘very’, ‘extremely’ and ‘not’ are also quite important. EVE’s AI recognises these words and their relation to one another. The answers are then given an overall sentiment – basically determining whether particular areas are doing well or need to be improved. From that it can be understood how employees are feeling and improvements can be driven on a ‘continuous listening’ basis.
Whether it’s chatbots or something more advanced, for the near future at least, it looks like AI will have a broad impact in attempting to improve the quality of our working (and personal) lives, and in making things more efficient.