When the Commonwealth Bank of Australia (CBA) announced in July that it was cutting 45 call centre jobs and replacing them with an AI chatbot, the move was positioned as a natural step in technological progress. Automating simple queries, the bank argued, would free up teams to handle more complex customer problems that require empathy and judgment.
Yet, just three weeks later, the bank was forced into an embarrassing backtrack. Facing rising call volumes and customer dissatisfaction, CBA admitted its decision was an “error” and apologised to affected employees. It reinstated roles after acknowledging it had failed to fully consider the business impact of removing human agents from customer service.
This incident underscores a critical reality: replacing people with AI in customer-facing roles is not as straightforward—or risk-free—as it seems.
Why AI Alone Struggles in Customer Service
Chatbots and voice-bots have been in use for years. They can handle routine FAQs, password resets, or transactional queries. But when customers face unique or urgent issues, AI often falls short.
I have personally felt stuck in endless loops with customer care chatbots with Airtel, Amazon India, and a few other notable brands. I’m not even sure if these were AI-driven or just rule-based non-Gen AI bots, but the frustration has been real. At Amazon India, for instance, reaching a live agent is unnecessarily painful. For now, I’ve found a workaround until they change their IVR algorithm, but that in itself tells you how broken the system feels.
Think about it: can a chatbot—even an advanced AI—really understand that I ordered 8 kg of detergent but received only 4 kg, possibly pilfered somewhere in the supply chain? Or how to register the frustration when a charging cable that was supposed to be inside a package was simply missing? These are everyday, nuanced customer problems that require a human ear, not a script.
The limitations are clear:
Confusion and loops: Chatbots often push customers in circles, with no real resolution. Lack of empathy: A human agent can gauge frustration, calm the customer, and offer practical solutions. Limited scope: Bots, no matter how “intelligent,” struggle with exceptions, missing items, or ambiguous cases.
When customer service is entirely devoid of human agents, these weaknesses become magnified. Over time, the outcome is predictable: frustration, eroded trust, and a spike in dissatisfaction.
The Industry Lesson: Balance, Not Replacement
CBA’s reversal highlights the importance of balance. AI should augment, not replace, human customer service:
AI for scale: Automating repetitive, transactional queries improves efficiency and lowers costs. Humans for impact: Skilled agents must remain available for complex, sensitive, or high-value interactions. Hybrid model: The most resilient organisations combine AI efficiency with human empathy, ensuring customers don’t feel abandoned in moments of need.
Industries such as banking, healthcare, insurance, and e-commerce—where customers often deal with high-stakes or emotionally charged issues—must be especially cautious. In these sectors, customer trust is fragile. A chatbot that can’t resolve a pressing problem is not just inconvenient; it damages the brand.
Looking Ahead
AI is here to stay in customer service, and rightly so. But the Commonwealth Bank’s misstep is a cautionary tale. Efficiency gains mean little if they come at the cost of customer loyalty.
For businesses, the key is not to ask “How many jobs can AI replace?” but rather “How can AI empower human agents to deliver better service?”
Because at the end of the day, technology should enable empathy—not eliminate it.
Have you ever had a frustrating experience with a chatbot or automated service? How did it affect your perception of the brand?
