Call Center

A call center is an operation where a great many people are on the phone at the same time. They take calls from customers or call customers themselves. Typical topics include complaints, orders, billing questions, or technical problems. Large companies usually don’t run such centers themselves but instead purchase the service from specialized firms. Worldwide, several million people work in this industry, with particularly many in India and the Philippines. It is precisely this industry that has been at the center of the debate about AI and jobs for several years now.

Why phone jobs of all things are coming under pressure first

Work in a call center is highly standardized. Many conversations follow a fixed sequence, the so-called script. The employee asks for the customer number, checks something in the system, and reads out a matching answer. A language model, meaning an AI system for text and speech, can replicate exactly these kinds of procedures. That’s why call centers are the first major testing ground for the commercial use of AI in customer contact.

On top of that comes economic pressure. Personnel is by far the largest cost block in a call center. A conversation with a human costs anywhere from a few cents to several euros, depending on the country. An AI response costs a fraction of that and gets cheaper every year. For companies, the math is therefore simple, even if the quality isn’t yet right everywhere.

For the financial world, this industry is an early indicator. Companies like Teleperformance or Concentrix are publicly traded and derive almost all their revenue from this work. When the first convincing AI assistants appeared in 2023, their share prices dropped sharply. Investors read from these prices how seriously the market takes automation.

From the hold-music menu to the talking assistant

You’ve known the simplest form of automation for a long time. A recorded announcement says: Press one for billing. Such menus have existed for decades, but they are rigid and often annoying. They can only do what someone has previously programmed in as a fixed sequence.

Modern systems work in three steps. First, speech recognition converts what is said into text. Then a language model processes this text and formulates a response. Finally, an artificial voice reads out the response. In doing so, the model is usually not allowed to invent freely, but instead draws on the actual customer database and internal manuals.

In practice, humans and machines often work together. The AI answers simple standard questions entirely on its own. For difficult or emotional cases, it hands off to a human. Another common application is the assistant working in the background: it listens in and displays suitable answers on the employee’s screen. Studies show that new employees in particular become noticeably faster as a result.

Where you encounter this in everyday life and in the news

When you call a bank, a mobile provider, or an online shop, you almost always end up at a call center. The chat on a website belongs to this category too, even though nobody speaks there. Experts therefore now mostly call such centers contact centers, because phone, email, and chat all flow together.

The term regularly appears in business news when companies justify job cuts. The Swedish payment service provider Klarna announced in 2024 that its AI assistant was doing the work of around 700 full-time employees. Later, the company partially backtracked and hired people again because quality suffered. Such reversals are typical of the current phase.

A common misconception is that AI completely replaces the profession. A shift is more realistic: simple inquiries disappear, and the remaining conversations become more difficult. Someone working in a call center therefore gets fewer routine questions and more complicated complaints more often. This changes the demands of the job more than pure employment figures show.