ChatGPT : Comment Rédiger un bon Prompt en 2026 ?

Item: ChatGPT : Comment Rédiger un bon Prompt en 2026 ?
Rating: 4
Author: PostSingularity Institute

Applied Sciences & Engineering AI — Models & Research UYQArtificial intelligence (AI)UYQFGenerative artificial intelligence

🎙 Ludo Salenne 👥 267K 📅 February 17, 2025 ⏱ 32 min 👁 135K 📄 tutorial 🧭 2026-08-21

Available in: English (current) Français

Keywords

promptChatGPTtokenizationenergyefficiency

Summary

The video is a tutorial by Ludo Salenne on how to write effective prompts for ChatGPT in 2025. It begins with a live comparison showing that a well-crafted prompt on a weaker model (GPT-4o) can outperform a vague prompt on a top-tier model (o1 pro). The creator explains the technical workings of LLMs, including tokenization, encoding, vectorization, and probability-based text generation, to justify why precise prompts lead to better and more energy-efficient responses. He highlights the environmental impact of AI, citing that a single request can consume significant energy, and introduces the French government’s ComparIA tool for comparing AI models’ energy consumption. The video also criticizes misleading advice on LinkedIn and presents a structured ‘perfect prompt’ framework, tested hundreds of times, which includes clear objectives, context, constraints, and examples. A promotional segment for his paid training course is included. The tutorial concludes with a disclaimer about the evolving nature of AI and a mention of OpenAI’s Prompt Generator tool.

159 words

Critical Evaluation

Value of the Information & Strength of the Argument

The video provides valuable, actionable information on prompt engineering, supported by live demonstrations and comparisons. The argumentation is solid, explaining the technical underpinnings of why prompt quality matters, and it effectively debunks common misconceptions. The emphasis on energy efficiency adds a novel and responsible angle, encouraging users to optimize prompts not only for better results but also for environmental sustainability.

Scientific Rigor, Source Quality, Title Accuracy

The video demonstrates scientific rigor by referencing official tools (OpenAI tokenizer, ComparIA) and a healthcare article on AI sustainability. However, some claims, such as specific energy consumption figures and model parameter counts, are presented without precise citations, relying on approximations. The title is accurate, though the ‘2026’ is slightly misleading as the content is focused on 2025. The creator’s promotional content is clearly separated from the educational content.

143 words

Title / Content Match

The title accurately reflects the content, which focuses on how to write effective prompts for ChatGPT in 2025, despite the '2026' in the title.

Quality & Reliability

7/10

The video provides a practical tutorial on prompt engineering, supported by live demonstrations and references to official tools (OpenAI tokenizer, French government's ComparIA). However, some claims about energy consumption and model parameters are approximate and lack precise citations.

Chapters

Cited Sources

ComparIA - French government AI comparison tool — Used to compare AI models' energy consumption and response quality.
OpenAI Tokenizer — Shown to demonstrate how prompts are tokenized.
Healthcare in Europe - GenAI LLM healthcare prompt sustainability — Referenced for statistics on energy consumption of AI requests.
OpenAI Playground — Mentioned as a tool for testing prompts.
Ludo Salenne's ChatGPT training course — Promotional link for his paid training.
Free AI resources (Prompts & GPTs) — Free resources for viewers.

Concurring Sources

OpenAI Tokenizer — Confirms the tokenization process described in the video.
ComparIA - French government AI comparison tool — Provides real-world data on AI energy consumption, supporting the video's claims.

Dissenting Sources

LinkedIn post by Michel Aguillard — The video criticizes advice from a LinkedIn post that suggests not assigning roles to AI, which the creator argues is based on a misunderstanding of LLM functioning.

External References

Contribution & Novelties

The video adds value by linking prompt engineering to energy efficiency, a perspective not commonly emphasized in mainstream tutorials. It provides a practical, tested framework for crafting prompts and debunks common myths. The live comparisons between models of different tiers are particularly illustrative.

Pour aller plus loin :

Prompt engineering - Wikipedia — Overview of techniques and best practices.
Tokenization (NLP) - Wikipedia — Explanation of the tokenization process.
Energy consumption of AI - Nature article — Research on the environmental impact of AI models.

85 words

Radar Profile

The radar profile shows high scores in information quantity and technical level, reflecting the video's comprehensive coverage of prompt engineering and its technical basis. The quality and reliability scores are slightly lower due to some unverified claims and promotional content.

Reliability 7/10

💬 Très positif. Sur les 30 commentaires analysés, les spectateurs expriment une forte approbation, saluant la clarté, la pédagogie et la pertinence du contenu, avec des remerciements récurrents pour les conseils pratiques et l'aspect environnemental.