Replacing workers with AI sounded spectacular. There are already companies backward

That an AI takes away our work is A fear That is being realized. More and more companies that bet on replacing human employees with automated processes with AI. At the same time, the potential that the AI agents It seems almost unlimited. The reality is much more complex and there are already some companies that have had to give reverse in their plans.

Yes, but to supervise her humans. There are many companies that have begun to draw a plan to replace their employees with AI, especially in the sector of the customer service. However, according to a survey of Gartner, 50% of companies are abandoning this path due to the challenges that the transition to a service managed only by AI is assuming. 95% of the executives surveyed claimed that they would bet on a hybrid approach combining AI and Human agents. In the words of Kathy Ross, director of the Gartner customer service: “IA offers significant potential to transform customer service, but it is not the panacea. The human touch remains irreplaceable in many interactions.”

The Klarna case. Perhaps the most popular case of a company that has regretted betting on the AI ​​is that of Klarna. In February 2024, his CEO presumed that an AI was doing the work of the 700 employees who had just fired and also assured that he was receiving the same score from the clients. A year later, the quality of service offered by AI received many criticisms and had to back to hire human labor Ensure that customers can always talk to a human agent.

Not so quickly. Although currently the agents of AI are not so reliable As Altman promisedit does not guarantee that they will not improve in the future. In addition, there are cases of other companies to replace humans for AI has gone well. Like the CEO of this Indian startup that He fired 90% of the template And a year later he states that it has been A success. And it is not the only one. Duolingo, for example, fired 10% of its translators to replace them with an AI. Other large companies such as UPS and Cisco They have followed similar paths. The case of IBM It is particular, since, although it has opted for AI, it has also had to hire more workers precisely to manage that AI.

IA agents. With the generative AI in decelerationsince the end of last year there is another concept that has gained relevance, that of the AI agents. While chatbots can only attend one request at the same time, an AI agent is able to perform more complex tasks. For example, you could organize the holidays only to mention the dates and destiny. The promise of autonomy, the ability to handle several tasks at the same time and make decisions make the AI ​​agents A threat to many jobs.

Unreliable. We can all make mistakes and in this it seems that AI agents are very human. The problem is that if an AF agent makes an error, he continues to repeat in successive tasks, making the error increasing And the final result is also wrong. We have also seen other uses of AI tools at work that have not fully went well, such as disastrous work interviews conducted with AI.

A revealing experiment. A experiment carried out by researchers from the Carnegie Mellon University He suggested that they are still very green. They created a fictitious software company whose employees were AI agents. They used Google, OpenAi, Anthropic and Meta models, to which they assigned roles such as finance, administration or software engineering. The result was disastrous: they only managed to complete 24% of the tasks assigned to them.

AI and labor market. The impact of AI on the labor market is undeniable. According to the last World Economic Forum Reportit is expected that by 2030 92 million jobs will be destroyed due to AI automation. But there is another face of the currency. At the same time, they will be created 170 million new positions And the AI ​​will be one of the job creation engines, something we have already seen with Salesforce’s case.

Cover image | Pixabay

In Xataka | The workers have stopped fear of AI as a machine to destroy jobs: software engineers do not think the same