How do I write a good AI image prompt if I'm not artistic?

You don't need to be artistic. You need to be specific. That's the whole trick. Think of it less like painting a picture and more like giving directions to a very literal, very fast intern who's never seen the world before. You wouldn't say "draw a dog." You'd say "a golden retriever sitting on a wooden dock at sunset, looking at the camera, photorealistic style." See the difference? The second one gives the AI something to work with. Most AI image tools — like Midjourney, DALL-E 3, or Adobe Firefly — respond best when you break your idea into four chunks: the subject, the action or pose, the setting, and the style. I've found that adding a style reference at the end makes the biggest difference. Words like "35mm film photography," "watercolor painting," or "3D render" steer the whole output. A concrete example: instead of "a robot in a city," try "a small rusty robot sitting on a park bench in a rainy futuristic Tokyo, neon lights reflecting in puddles, cinematic lighting, photorealistic." That prompt will give you something usable on the first try. A common mistake beginners make is overloading the prompt with conflicting ideas. Keep it to one clear scene. You can always generate variations. Also, don't ignore the negative prompt field if the tool offers one. Telling the AI what you *don't* want — "no blur, no extra limbs, no text" — is just as helpful as telling it what you do want. It's like putting up guardrails. The real insight here is that prompt writing is iterative. Your first result will rarely be perfect. Tweak one thing at a time and regenerate. That's the actual workflow professionals use.