AI in Customer Service: A Double-Edged Sword
The recent Air Canada incident, where a customer was misled by a chatbot about bereavement discounts, highlights the growing concerns surrounding AI-powered customer service. This case exemplifies the concept of “botshit” – incorrect or fabricated information produced by chatbots. As businesses increasingly adopt AI technologies, the need for critical management of chatbot-generated responses becomes paramount.
Key Points on AI in Customer Service
- Generative AI use in workplaces has nearly doubled in the past six months
- 79% of business leaders believe AI adoption is crucial for competitiveness
- Chatbots can lead to situations where no employee is accountable for algorithmic errors
- Recent studies show high inaccuracy rates in AI-generated responses, particularly in specialized fields
The Balancing Act: Efficiency vs. Accuracy
While AI chatbots offer streamlined decision-making and improved efficiency, they also present risks to customer experience, company reputation, and legal liability. The challenge lies in harnessing AI’s potential while mitigating associated risks. Experts suggest implementing rigorous guidelines, guardrails, and restrictions on AI chatbots’ vocabulary and response topics. Additionally, practice-specific chatbots using retrieval augmented generation show promise in reducing inaccuracies.
