← Retour au blog
tech 12 June 2026

Our Workplace LLM Mass Delusion

In the current AI hype, language models (LLMs) are often seen as the miracle solution. However, the adoption of these technologies in companies reveals unexpected challenges and unfulfilled promises.

Article inspired by the original source
our workplace LLM mass delusion ↗ blog.avas.space

Introduction

The enthusiasm for language models (LLMs) like ChatGPT or Copilot dominates discussions on automation and artificial intelligence in the workplace. However, this fascination might turn out to be a collective delusion, particularly in work environments struggling to find real added value in these technologies.

The Promise of LLMs

LLMs promise to automate repetitive tasks, improve productivity, and reduce costs. In theory, these tools should free up time for higher-value tasks. According to a McKinsey study, automation could lead to up to 20% cost savings in certain sectors. Yet, this promise doesn't always materialize.

The Reality of Business Adoption

In many companies, LLMs are eagerly adopted, but the results often fall short of expectations. Take the example from Ava's blog: despite significant investments in consultants, workshops, and licenses, LLM-based projects struggle to demonstrate their effectiveness. Employees often face increased complexity and unreliable results.

Challenges with LLMs

Accuracy and Relevance Issues

LLMs, while powerful, are not infallible. They can generate errors, hallucinate information, or fail to understand the specific context of a task. A Gartner study highlights that 54% of companies adopting AI have encountered accuracy issues with their models.

Hidden Costs

Adopting LLMs involves often underestimated costs: training employees, adjusting processes, and managing errors. Additionally, licenses can be costly for companies already under financial pressure.

Recurring Failure Cases

Ava's company is not an isolated case. Many organizations report similar experiences where LLM projects fail to generate productivity gains. Internal meetings often reveal that the tools struggle to integrate effectively into existing workflows.

Lessons Learned

Adopting a Realistic Approach

To leverage LLMs, companies must adjust their expectations. Rather than seeking a miracle solution, they should focus on specific use cases where LLMs can truly add value.

Investing in Training

Training teams to understand and properly use LLMs is crucial. This includes developing 'prompt-engineering' skills and having a clear understanding of the current technology's limitations.

Conclusion

The fascination with LLMs in workplaces may well be a collective delusion, but it should not overshadow the real potential of these technologies when used wisely. For decision-makers, it is crucial to navigate prudently and pragmatically in this constantly evolving technological landscape. Let's discuss your project in 15 minutes.

LLM automatisation IA productivité formation
Deepthix newsletter · 100% AI · every Monday 8am

An AI agent reads tech for you.

Our AI agent scans ~200 sources per week and ships the best articles to your inbox Monday 8am. Free. One click to unsubscribe.

Visit the newsletter page →

Want to automate your operations?

Let's talk about your project in 15 minutes.

Book a call