When AI Replaces Humans… Then Backtracks

When AI Replaces Humans… Then Backtracks

‍

Artificial intelligence promises to transform businesses: automating tasks, reducing costs, and improving productivity. On paper, it can quickly process simple requests, operate 24/7, and manage large volumes of data.

But in reality, AI also has its limitations: it lacks empathy, struggles with complex situations, and can degrade the customer experience. It is precisely for these reasons that some companies that replaced humans with AI ultimately had to reverse course.

‍Klarna: AI-Driven Customer Service Degrades Customer Relations

‍

The Swedish fintech company Klarna was one of the most striking examples. In 2024, the company decided to invest heavily in AI for its customer service. It eliminated approximately 1,800 positions and implemented a chatbot capable of handling millions of requests, resulting in significant cost savings.

However, the limitations quickly became apparent: customers complained about impersonal, sometimes inappropriate responses, and the inability to speak to a human. Service quality plummeted, and the brand image suffered.

Faced with these criticisms, Klarna changed its strategy in 2025 and rehired human staff. The company now adopts a hybrid model: AI for simple tasks, humans for complex cases.

‍The problem: AI is effective at processing simple requests, but fails as soon as it needs to understand emotions or solve specific situations.

‍McDonald’s: Automated Orders… But Imperfect

‍

McDonald’s also tested AI in its drive-thrus, notably with automated voice ordering systems. The goal was to partially replace employees to speed up service.

However, several incidents were reported: incorrect orders, misunderstandings with customers, and even absurd situations that went viral on social media. These errors damaged the customer experience and the brand image.

As a result, McDonald’s scaled back or stopped some of its experiments, returning to more controlled solutions.

‍The problem: AI struggles with natural language, accents, and unexpected situations.

‍Commonwealth Bank of Australia: The Limits of Banking Chatbots

‍

Commonwealth Bank of Australia also experimented with using chatbots to replace part of its customer service. As in other cases, the goal was to reduce costs and automate interactions.

However, these systems have shown their limitations: lack of reliability, errors in responses, and an inability to handle sensitive or complex requests. Some banks have had to bring back human staff to mitigate risks and maintain customer trust (especially in a sector as critical as finance).

‍The problem: in sensitive areas like banking, errors are unacceptable, and human intervention remains essential.

‍Conclusion: Towards a Hybrid Model

‍

‍These examples illustrate an important point: AI does not completely replace humans, it complements them.

‍It is very effective for:

• repetitive tasks
• simple requests
• rapid data processing

‍But it remains limited for:

• human interaction
• managing emotions
• complex or sensitive cases

This is why companies are now moving towards a hybrid model, combining artificial intelligence and human intervention. The challenge is no longer to replace humans, but to find the right balance.

‍Sources:

• Forbes
• Science&Vie
• Le journal de la Finance 
• WebProNews

‍