Transcription
Tesla, Microsoft, Amazon, Google, and almost every tech company you can imagine have tried to replace humans with AI.
This week, Microsoft announced it is laying off nearly 4% of its workforce. The CEO of Amazon told the company's employees that some of their jobs will be replaced by AI. In recent years, companies have laid off huge numbers of workers. The new year is not starting off on a happy note for workers at some of the country's largest tech companies. This was because companies were overconfident about their AI and automation developments. Amazon is warning its employees that the company will have a smaller workforce in the future because of generative AI and agents.
However, it seems that companies are quietly backpedaling as according to an org view of senior business leaders and executives, 55% of businesses that replaced employees with AI regret it. Tesla, CLA, IBM, Duolingo, and countless more. They were overconfident in replacing humans. But now most are quietly backpedaling. So why is tech regretting replacing humans with AI?
Company's attempts to automate processes with minimal human intervention are not new. Tesla's case in 2017 is an early example of the limits of extreme automation. The company sought to build the so-called machine that builds the machine with a production workshop almost entirely automated for the Tesla Model 3. Elon Musk, Tesla's co-founder and CEO, has referred to the Gigafactory as the machine that builds the machine. And it's all part of his master plan to make electric cars more affordable. The goal was to achieve the production of 5,000 vehicles per week, but results were significantly lower due to machinery failures, production bottlenecks, and systems that did not operate as intended. Robots specialized in tasks such as placing front seats broke down up to five times a day, while industry standards show operational intervals exceeding a month without interruptions. The combination of new technologies and lack of preventive maintenance turned the production line into a fragile system. Tesla had to reintegrate human staff through a temporary line called the Sprung project where workers replaced the robots. In April, Musk tweeted that, "Humans are underrated." He was referring to Tesla's experience making its new Model 3s with one of the most robotics dependent assembly lines on the planet. That grand experiment failed to deliver nearly the number of cars Tesla promised. Production eventually accelerated, allowing the company to avoid bankruptcy. But this example shows that the efficiency promised by automation can generate organizational fragility when the value of human labor is underestimated.
With the adoption of advanced language models, companies began applying this logic to the replacement of intellectual and service tasks, including customer service, marketing, data analysis, and content development. CLA, for example, implemented chatbots that replaced hundreds of agents, reducing its workforce from 5,000 to 2,000 employees. Initially, the company reported that chatbots managed 2/3 of interactions, but later a decrease in service quality was identified. In 2023, the CEO of CLA said AI would replace half his workforce in just a few years. And CLA eventually reduced its workforce by 40%. But in 2025, Sebastian flipped that narrative on its head. He publicly stated that quality human support is the way of the future for us. In fact, leaked internal data indicate that problem resolution times increased by 27% while unsatisfactory interactions grew by 35% in the first 3 months after implementation. Failures included incorrect approval of leave requests and inadequate responses to internal conflicts. This showed that AI is more effective in structured tasks than in processes requiring contextual judgment.
Similarly, Taco Bell experimented with an automated voice system in 500 locations. However, the company had to limit its implementation due to errors in orders and billing which affected customer experience and operational efficiency. I don't know if either of you two have interacted with an AI chatbot customer service thing, but it is a nightmare. I was stuck in like a chatbot death loop of like trying to get it to do what I want. Eventually, I had to, you know what, pick up a phone and call a human being.
Duolingo implemented its system called AI first to replace part of contractor's work aiming to improve efficiency and reduce costs. However, the company later reported a noticeable decline in lesson quality with errors affecting up to 42% of content in some courses and causing an 18% drop in user retention during the first quarter after the changes. Another example is the Australian company Telstra, which replaced 2,800 employees with AI, but saw customer response times increase by up to 25%. Shopify conditioned the hiring of new employees on proving that AI could not perform certain tasks, generating project delays and an internal environment of uncertainty. MIT analyses indicate that only 5% of AI integrations generate immediate revenue increases. The remaining 95% fail to achieve significant results mainly due to insufficient planning, lack of adequate data, and inadequate employee training in the use of these tools. Tech giants like Microsoft and Google are outsourcing more and more coding to AI in a productivity push, but some new research shows the tools might not be as helpful as some expect.
Organizational impact is also reflected in employee turnover. Companies that implemented AI without human supervision experienced an average increase of 22% in voluntary turnover during the first 6 months, raising recruitment and training costs by 18%. Additionally, customer experience was affected with decreased satisfaction and loyalty metrics. The fragility of automation is evident when a small AI breakdown can halt entire operations while human intervention can resolve incidents immediately. Despite these challenges, AI can be complimentary if implemented strategically. 80% of leaders plan to train their employees in AI tools and 41% have increased their learning and development budgets. And we felt that it was really important um that we help upskill people. Uh so we we are investing a lot into this.
Startups and companies that apply AI gradually and purposefully report productivity increases of up to 35% and operational cost reductions of 27%. This shows that the combination of AI with human supervision produces better results. In logistics, for example, route optimization with AI accompanied by human supervision has reduced delivery delays by 18% without compromising customer experience. But the implementation of AI in companies has shown mixed results in terms of employee turnover. On one hand, automating repetitive tasks can free employees for more strategic roles, reducing burnout and improving retention. On the other hand, the perception that AI can replace jobs can generate job insecurity and increase turnover. A new article from Business Insider describes so-called office paranoia, and it outlines how factors like artificial intelligence and changes to the labor market are bringing a sense of dread to workplaces all over the place.
Furthermore, leaders recognize that AI implementation can generate financial and operational benefits. However, its effectiveness depends on comprehensive planning that considers training, supervision, quality protocols, and adaptation to existing processes. AI is also proving not to be as effective on its own. A MIT study revealed that only 7% of AI initiatives in companies generate a significant return while the remaining 93% produce no measurable results. This finding underscores the importance of strategic and well-planned AI implementation considering the specific needs of the company and proper integration with existing processes. Evidence suggests that AI works best when it complements human skills rather than attempting to fully replace them. This approach allows employees to focus on higher value strategic tasks while automated systems handle repetitive or large-scale analytical functions.
A new poll shows more workers in the US are using artificial intelligence to help free up hours at work. The survey from Resume Builder shows 40% of people using ChatGPT say they save 1 to 5 hours per week. Data also show that internal perception of AI affects adoption and results. When employees perceive that their work is devalued, the risk of stress, demotivation, and talent loss increases. Therefore, responsible AI integration requires clear communication about objectives, benefits, and limitations of the technology as well as continuous training programs. Evidence from companies such as Tesla, CLA, and IBM demonstrates that organizational resilience depends on the balance between technology and human capacity.
[Music]
At Economy Media, your opinion matters to us.