We give extra weight to expert or high-status opinions, even when they may be wrong.
More quotesPrompt one: “Give me ideas for a product launch.”
Output: teaser campaign, countdown, influencer partnership, interactive experience, exclusive preview.
Every noun has arrived from the statistical centre of a thousand tired requests. Nothing is wrong enough to delete. Nothing is specific enough to use.
Prompt two adds facts. Buyers are suspicious engineers. Budget is fixed. The product replaces a manual check people secretly enjoy. Every idea must demonstrate one provable advantage in thirty seconds. Avoid countdowns, influencers and claims about revolution.
Output one proposes filming the manual check beside the automated result.
Output two proposes asking engineers to find the discrepancy.
Output three invents a result absent from the supplied evidence.
One generic answer has disappeared. One usable mechanism has appeared. One confident lie has also appeared.
Prompt three asks for ten variations on the discrepancy demonstration, each using only the supplied test result. Seven are cosmetic cousins. Two introduce evidence the model was never given. One turns the disliked manual step into the comparison.
The persona changes from “world-class creative” to “cautious buyer trying to reject this claim”. Objections become more specific. Authority remains simulated.
Prompt four requests a circus clown. The copy acquires juggling.
Prompt five removes the persona and includes three customer quotations. The response borrows the phrase “I still check it myself”.
The final prompt is saved with the source material attached. The next response begins generating at 14.06.