← Retour au blog
tech 26 June 2026

The Exhaustion of Talking to a Tool

Language models, while useful, deplete our social energy without offering human reciprocity. Are they truly the best use of our mental resources?

Article inspired by the original source
The Exhaustion of Talking to a Tool ↗ ohadravid.github.io

Introduction

Talking to language models has become a routine for many tech professionals. However, this interaction can be surprisingly exhausting. Why? Because using these tools demands a social energy expenditure that might be better spent elsewhere.

The Magic of Tools

When you use a good tool, your brain adapts and considers it an extension of your body. Think of the fluidity with which you drive a car or type on a keyboard. It's intuitive, natural. With language models, it's a different story. They require constant communication, much like engaging in a social conversation.

The Social Cost of LLMs

Language models, like GPT-3 or Claude, aren't fast or consistent enough to seamlessly blend into our daily lives like a keyboard does. They demand from us a form of "social tax": negotiating, convincing, sometimes even getting angry at a tool meant to make life easier. According to a recent Gartner study, 70% of professionals find LLMs lack intuitive responsiveness, complicating their adoption.

Comparison with Human Interactions

Interacting with people offers rewards that LLMs cannot match: mutual learning, inspiration, collaboration. A survey conducted by McKinsey shows that 85% of employees feel more valued and inspired after a productive meeting with their peers than during a session with an AI tool.

When Are LLMs Useful?

Of course, there are tasks where LLMs are invaluable. For repetitive tasks or large-scale data analysis, these tools outperform human capabilities. A Forrester report indicates that companies integrating LLMs into their software testing processes have seen a 30% reduction in human errors.

Conclusion

While language models can accomplish wonders, it is crucial to ask whether the social energy spent is worth it. For some tasks, certainly. But for all? Not necessarily. Directing this energy towards human interactions could be far more beneficial.

Let's discuss your project in 15 minutes.

IA modèles de langage énergie sociale productivité interactions humaines
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