10 ChatGPT prompt tricks I use - to get the best results, faster (www.zdnet.com)

ZDNET lays out ten practical prompt-writing techniques to get faster, more accurate results from ChatGPT. The piece stresses conversational, “human” prompts and interactive multi-step questioning—treat the model like a coworker, iterate on answers, and feed follow-ups to refine output. It highlights role-play and persona prompts (ask the model to be a product manager, teacher, pirate, etc.) and audience framing (“explain X to a non‑technical board member”) to control tone, depth and perspective. The author also recommends constraints (word limits, format), explicit context, and asking the model to justify claims or cite evidence to reduce hallucinations and keep responses honest. Technically, these tips map to common prompt‑engineering patterns: conditioning the model with context and identity, using chain‑of‑thought style follow-ups to surface reasoning, and applying human‑in‑the‑loop iterations for quality control. A notable practical detail: the author limits creative outputs to 500 words because they observed reliability issues when asking for 500–700 words—an example of working around current LLM generation quirks. For AI/ML practitioners and power users, these tactics improve reproducibility, reduce error-prone outputs, and make LLMs more useful across tasks from technical explanations to creative writing.