Chapter 2

HOW TO USE GENERATIVE AI

Generative AI is no longer science fiction: today it creates texts, images and even code in seconds, but how can we make the most of it ?

Generative AI is a system (e.g., ChatGPT, DALL·E, Midjourney) that creates new content: text, images, audio, or code. To use it effectively:
Define your goal : What do you want to achieve (persuasive copy, product image, technical summary)? Be specific.
Choose the right tool : text templates for copy/ideas, image templates for visuals, code templates for programming.
Provide quality input : context, examples, desired output format. The AI ​​repeats what you give it.
Refine and iterate : generate multiple versions, adjust the prompt, and blend outputs to achieve the best.
Review and edit : correct errors, verify facts, adjust style and human voice.
Protect sensitive data : do not upload private or confidential information carelessly.

Prompt engineering: is the art of writing instructions that guide the AI ​​toward the expected result. Key techniques:
Be specific : length, tone, audience, format.
Provide context : project brief, examples of desired output.
Assign a role : “Act as… (expert, copywriter, designer).”}
Ask for structure : “Deliver in 3 bullets, headline, and CTA.”
Iterate : Refine based on the previous output.
Useful templates (quick): Copy (Instagram)
You are a copywriter. Write 3 Instagram captions (max. 50 words each) about a reusable water bottle, approachable and optimistic tone, audience: eco-conscious millennials. Include 2 hashtags and a call to action.
Product description: Act as a senior copywriter. Write a 120–150-word product description for an online store explaining the benefits, materials, and three technical features. End with a CTA.
Prompt image (Midjourney / DALL·E): Advertising photo of a stainless steel water bottle on a wooden table, forest background at sunset, warm lighting, minimalist composition, editorial style, high resolution, no text.
Code (generation): You are a Python developer. Write a function, calculate weighted average (list, weights) with a docstring and 3 unit tests using pytest.
Summary: Extract the following text into 5 bullets with the main ideas and a practical recommendation sentence.

Beyond chatbots: generative AI does much more than just chat:
Content creation : articles, scripts, emails, posts.
Design and prototyping : renders, mockups, mood boards.
Programming : code templates, refactoring, testing.
Audio and video : synthetic voices, scripts, assisted editing.
Research : summaries, source comparison, data mining.
Automation : report generation, scalable templates, mass customization.

When and how to use generative AI. Use it when :
You need to generate many quick ideas or drafts.
You want to prototype visuals or text for A/B testing.
You're looking to speed up repetitive tasks (summaries, labels, first drafts).

Avoid or exercise caution when :
Critical accuracy is required  (legal, medical, sensitive data) without human review.
Private information is involved  that cannot be shared.
The output has unfiltered ethical or social impact (misinformation, bias).

Best practices:

Human-in-the-loop : Always have an expert reviewer.
Fact-check : Confirm important data.
Version control : Save prompts and output for traceability.
Mitigate bias : Review language and representations.
Transparency : If applicable, indicate that the content was assisted by AI.
Quick pre-publish checklist:
Did I check the facts?
Is the voice/tone correct? Is there a privacy or copyright risk?
Are metadata and references clear?
Was there a human review?

Tools in practice: case study (product launch)
Scenario: Launch of an eco-friendly reusable bottle.
Objective : Web description, 3 Instagram captions, 3 visual creatives.
Tools : ChatGPT (copy), Midjourney/DALL·E (images), Figma (layout).

Workflow :
Research: product brief + audience.
Ideation (IA): Generate 10 titles, 12 captions, 6 descriptions.
Selection: The team chooses 3 options per type.
Design: Use prompts for product renders.
QA: Factual verification, tone review, A/B testing.

Launch and measurement.
Prompt examples for this case
. Web description:
Act as an e-commerce copywriter. Describe the “EcoFlow 500” bottle in 130 words, highlighting the material, durability, plastic alternative, and a 2-year warranty. Include a final sentence as a call to action.
Instagram caption: Write 3 different captions (max. 40 words) for millennials interested in sustainability. Tones: 1) funny, 2) emotional, 3) informative. Add 2 relevant hashtags and a CTA.
Image prompt: Product photography: EcoFlow 500 bottle on a rock by a river, sunrise light, visible water droplets, sharp focus, editorial style, high resolution, centered composition.
Metrics to evaluate:
CTR and conversion on the product page. Engagement rate (likes, comments, shares) on social media.
Return on creative investment (time saved vs. performance).

Common mistakes and how to avoid them:
Vague prompts → ask for format, length, and examples.
Blind trust → always check and verify.
Forgetting the audience → define your voice and audience.
Lack of version control → save prompts and results.

AI Opinion: I've reviewed your article "How to Use Generative AI" and here's my detailed feedback:  
Strengths: Clear, practical approach : It doesn't get stuck in theory, but offers concrete steps for using generative AI effectively (defining your goal, choosing a tool, refining prompts, reviewing).
Pedagogical structure: It's organized in a progressive manner—what it is, how to use it, prompt engineering techniques, case studies, and a checklist—making it easy for any reader to understand. Applied examples: The
eco-friendly bottle case  is excellent because it shows the real-life use of different tools (ChatGPT, Midjourney/DALL·E, Figma) in a complete workflow.
Emphasis on accountability: You highlight good practices such as human review, data privacy, bias mitigation, and transparency, which lends a sense of responsibility to the article.
Accessible yet technical language: It's written in a way that can be useful to both marketing professionals and educators or beginners in generative AI.
Opportunities for improvement (optional)
: A more narrative introduction: You could start with a phrase that sparks interest, for example:  "Generative AI is no longer science fiction: today it creates text, images, and even code in seconds, but how can we make the most of it?" Concept
visualization: Adding tables or boxes with "common mistakes" or a "quick checklist" would help in a magazine-style version or visual guide.
Emphasis on ethical risks: You already mention them, but you could expand a bit on the impact of biases, fake news, or deepfakes.
Expand metrics: In addition to CTR and engagement, including efficiency metrics (hours saved, reduction in creative costs) would better show the business value.
My conclusion : The article is very solid, comprehensive, and action-oriented. It could work as:
A practical manual for companies that want to get started with generative AI; an educational guide for innovation, communication, or marketing workshops; and an informative publication in magazine format or an interactive PDF with visual examples of prompts and outputs.