Part 3 of The European Chatbot and Conversational AI Summit 2023–3rd Edition series
Our last two (1, 2) articles have respectively focused on the theoretical approach on conversational AIs and their business application scenarios. The last article of the three-part series is on chatbot best practices that help businesses move forward.
For this post, we’ve compiled three panel discussion sessions encompassing industry leaders’ and academic experts’ unique viewpoints on customer experience and current trends in conversational AI.
AI Everywhere: Perspectives on AI-First Customer Experience Strategies
The Growing Use of AI in Customer Service
The use of AI in customer service is becoming increasingly prevalent, and experts predict that AI-mediated channels will become the first point of contact for every use case across every channel in time. However, there are challenges in scaling up conversational AI systems in companies, and maintaining a good consistent experience for the user over time is a key concern.
During the session, the panel of experts discussed the future of AI in customer service. Serkan Ibrahim, VP at the European division of Kore.ai, believes that AI will handle most first contacts and normal day-to-day interactions, but a human will still be required in more adverse situations that demand real human attention.
Bev Bartlett, senior manager of TOBi Everywhere at Vodafone, believes that companies are purposefully heading towards an era where chatbots are as good and trustworthy as possible, so they can be everywhere in the customer service system. Santosh Kumar, lead of commercial service at Roche, agrees that conversational AI for customer service is now an obligation for big corporations, as AI facilitates in-depth understanding of their clients.
The Challenges of Scaling up Conversational AI in Companies
Eugene Neale, director of business IT and CX engineering at loveholidays, notes that chatbots like Sandy of loveholidays are becoming a central part of organizations. However, the challenge lies in being unaware of where to concentrate efforts to make the bot more broadly applicable. The panel discussed the main challenges of scaling up the conversational AI idea in a company, such as not going “too fast, too early” as discussed by Bartlett, and knowing where to stop integrating new technologies and techniques.
Maintaining Consistency and Trust in Conversational AI Systems
Maintaining a good consistent experience for the user over time is crucial for chatbot best practices. To achieve this, Eugene explains that loveholidays uses anomaly monitoring to detect if the bot’s conversations are deviating too much from what they consider as normal. Meanwhile, Bartlett emphasizes the importance of not adding new features to the system unnecessarily when the consumer is “already getting what they want.”
Conclusion: The Future of AI in Customer Service
To conclude, the experts believe that AI-mediated channels will be even more crucial in customer service as they eventually take place as the initial contact point for much of the consumer use cases. Taking this into account, companies must consider the challenges of maintaining a good consistent experience for the user over time when scaling up their conversational systems. Companies can maintain effective and trustworthy chatbots by implementing chatbot best practices like anomaly monitoring and avoiding adding unnecessary features.
Conversational AI, What is in there for the Customer?
The use of conversational AI in customer service has become increasingly common over the years. Many companies have already started using chatbots to handle their customer services, and the results have been positive.
But questions like ‘What is in it for the customer? How does conversational AI benefit them?’ remain. In this panel discussion, industry professionals shared the important points to consider when conceptualizing the potential consumer experiences through chatbot best practices.
Charismatic Bots for Extraordinary Experiences
Christoph Esslinger, the founder and managing director of VUI Agency, emphasizes the importance of having charismatic bots handling customer services. According to Esslinger, “Putting customers at the center of AI development is critical in providing an extraordinary experience, which not only cuts costs for companies, but also generates income.” He also believes that a bot’s good will can have a positive impact on a company’s image as a whole.
Instantaneous Interaction
Sometimes, users do not have the patience to wait for human assistance. Conversational AI can provide customers with the facility to speak with instantaneous machine support and receive information in a kind, helpful manner. This benefit saves customers time and enhances their experience.
Start with a Simple Chatbot
As a best practice for chatbots in businesses, Pooja Sudarshan Deshmukh, the vice president in digital transformation at Citi Bank, recommends beginning with a simple chatbot that can handle a limited range of questions.
In her experience working in the banking industry, Deshmukh suggests starting with basic queries such as “What products does this bank offer?” and “How can customers access these services?” By taking this approach, businesses can quickly develop a functional chatbot that is easier to manage and troubleshoot.
As the chatbot becomes more sophisticated, it can be expanded to handle more complex queries and provide a wider range of services. Overall, Deshmukh’s advice highlights the importance of chatbot best practices where the bot should start small and gradually scale over time.
Analytics and Feedback for Better Quality
Another crucial aspect in chatbot best practices is that teams should focus on their analytics during development.
However, when in production, the focus should drift towards qualitative feedback. Ruth Engemann, manager of intelligent automation at Deloitte Consulting, suggests that teams get feedback from users’ conversations with the bot and analyze details such as the time it takes for the user to exit the conversation, as well as the starting point, and the exit point of the conversation. If nearly all users leave the conversation as soon as they receive the response to the greeting message, it could be a clear signal that there is something wrong with the bot.
Another useful way to get indirect feedback from conversations is to train a model to detect user sentiment. This feedback provides insights into the user’s attitude towards the bot and helps teams improve its quality.
Experimentation with Customers
Jeroen Das, product owner of conversational AI and chat at ABN AMRO Bank, suggests that experimenting with customers is another way to improve the bot’s quality. By calling out some customers to come to a room and experiment with the chatbot, teams can receive qualitative feedback on how the model is performing.
Voice Recognition for Enhanced Accessibility
Engemann highlights that the advancement of bots that can understand human voice is a step forward for facilitating contact between the user and the company. Additionally, voice recognition models could benefit countries with low literacy rates where their primary means of mobile communication is through voice.
Contextual Understanding
One pertinent question raised by the experts is whether a chatbot could differentiate the context in which its user is when they begin the conversation. Accurately determining the context would allow the bot to convey things like humor and compassion without the fear of offending anyone.
To sum this session, we could see that the conversational AI benefits customers in many ways, including instantaneous interaction, and enhanced accessibility. By focusing on customer experience and experimenting with users, companies can continue to improve the quality of their chatbots and offer their customers an extraordinary experience.
The Next Wave of Conversational AI: Trends and Challenges
Not only can chatbots assist us with tasks and provide personalized recommendations, but they can also serve as virtual companions and even friends.
In this panel discussion, researchers and academic leaders shared their analysis on the current trends on conversational chatbot interfaces and the challenges they face.
Ensuring Safe and Accurate Responses
According to experts like Dr. Kulvinder Panesar, assistant professor of applied artificial intelligence at the University of Bradford, chatbots will play an integral role in the future of conversational AI.
Meanwhile, one of the key challenges facing conversational AI is ensuring safe and accurate responses from chatbots. Michael McTear, emeritus professor at Ulster University, suggests mitigating the risks of gross mistakes such as hallucinations and generating insulting outputs. To achieve this, multiple models can be used to improve intent matching and reduce errors. Additionally, advances in natural language processing and machine learning can help chatbots better understand and respond to user inputs.
Humanizing AI
As conversational AI continues to become more integrated into our daily lives, efforts to humanize it are becoming increasingly important. Lotte Willemsen, a professor (‘lector’) at Rotterdam University of Applied Sciences, emphasizes the importance of establishing trust with users and making chatbots more relatable. One approach to achieving this is through the use of digital humans, which are non-living human personas. Lotte discusses the potential of this technology in sensitive situations where individuals may fear judgment, such as seeking help for sexual diseases. Additionally, Lotte believes that digital doubles, which are essentially digital clones of oneself that can perform various tasks on one’s behalf, will become more prevalent in the future. These tools can enhance the humanization of AI and improve the overall user experience.
Advancing Dialogue
To further enhance the capabilities of conversational AI, experts like Angus Addlesee, research associate for conversational artificial intelligence at Heriot-Watt University, are calling for advancements in dialogue completion and multi-party dialogues. This requires AI to be able to understand context, complete sentences, and engage in conversations with multiple parties at once. Advancements in natural language processing, machine learning, and contextual understanding can help achieve this.
Addressing Current and Future Challenges
As with any technology, there are challenges facing conversational AI today and in the future. These include addressing issues such as human failure, data collection, and accuracy and safety of responses to images. To overcome these challenges, a user-centric approach to development is crucial, as is the need to continuously improve and evolve AI models through ongoing research and development.
To summarize this panel discussion, the future of conversational AI is bright, with chatbots serving as virtual assistants, companions, and friends.
However, challenges remain, such as ensuring safe and accurate responses, advancing dialogue capabilities, and addressing current and future challenges such as human failure and data collection. Adhering to chatbot best practices, as well as the continuous development in the field of natural language processing, machine learning, and contextual understanding will help overcome these challenges and unlock the full potential of conversational AI.