What are the benefits and challenges of using AI chatbots for customer service?

Chatbots for customer service have come a long way, but there is still work to be done. As someone who worked on a chatbot for software developers, I understand the challenges of creating an effective chatbot. One major issue is the impatience of customers who expect immediate results. Think of every time Alexa tells you the weather instead of shutting the lights off like you asked. However, as AI technology improves, chatbots are becoming more reliable and efficient. In the very near future, it's possible that people will look forward to interacting with chatbots on websites or online because they know they will receive excellent service. Although we're not quite there yet, the future for chatbots in customer service looks promising.

Having advised on and delivered dozens of contact center optimization programs using AI chatbots, here are the two sides of the coin Works very effectively for both the customer and the organization when the customer intent is very clear and the action required is very simple and the action required is basic (eg update my address, provide balance outstanding, email me a claims form.....etc) Needs a lot of work training the chatbots when the customer intent is nuanced and the action is quite ambiguous (eg explain why my premium has gone up 20%) How does one deal with it? Create an effective Triage model upfront using NLP algorithms that can accurately spot the intent so that complex intents are farmed out to a human

From a UX Designer's perspective, an AI chatbot is a valuable tool in the pocket that serves both - the customer by enhancing the experience, as well as the business. Let's take a look at several poignant examples: - comparing the 24/7 availability, unlimited by time zones or business hours AI chatbot with typical business hours, - delivers instantaneous responses, while the human counterpart has a queue and a waiting time to respond, - adeptly manages high volumes of customer inquiries while unearthing data-driven insights from user interactions and behavior patterns, and offers a wealth of insights for strategic decision-making, - ensures a uniform brand tone and style in standardized responses (e.g. seamlessly achieved in the Flo app).

In my experience which has only been since November 2022. I’ve seen the pros and cons. As a job seeker it’s helped me with mock interviews and I see the lack of the human touch. I still needed to revise many parts of my content to add that personalized element but it definitely speeds up the process. When it came to drafting articles, tailored cover letters, and career summaries I feel more confident with my work and content. It’s a big support to me and makes my work look more refined and professional.

Irrespective of technical intricacies that cause the malperformance of the bot, customers can initially be revolted when their queries fail to reach a real person, causing an immediate loss of personal touch. Furthermore, they must embark on a learning curve to effectively communicate with the chatbot as the bot needs simple wording, so putting the problem in large blocks of text may present itself as useless as the bot will have a hard time understanding the intent. Empathy often falls short in the chatbot's design and linguistic capacity, leaving customers feeling misunderstood. Privacy concerns come to the forefront when the users feel a reluctance and distrust to share data with a chatbot versus a human counterpart.

An AI Chatbot can likely be tuned to yield fuzzy results to simulate context clues/nuance, but every once in a while, customers will be misinformed or be unable to accurately describe what they need from Support. And in those cases, tribal knowledge, customer service experience, or simply prior run-ins with the same mistakes from other customers will be the key piece of information that leads to a timely resolution.

Depending on the overarching business goal, a Pareto analysis would be a good starting point when looking for e. g. scaling potentials. Distilling data to determine which customer inquiries have the biggest impact, finding relevant data that measure the impact, and developing features that address those inquiries. As for the 5 Why and Ishikawa diagrams, they are standard parts of the 8D Report of customer complaints in the automotive industry. In my experience, the challenge is that they all take an interdisciplinary approach, allocating resources. This is why it would be interesting to see the customer going through the 5 Why with the AI chatbot and at the end the customer would get a set of possible solutions and measures.

AI definitely has benefits in some industries. In my professional opinion though, there will never be a replacement for human eyes and human thinking/problem-solving. Knowing and understanding how to professionally and effectively communicate is a key part of career and organizational growth.