AI Reveals Gaps in Questioning and Prompting Skills
PlainEnglish published a July 7, 2026 essay arguing that generative AI exposes gaps in how people frame questions, specify outcomes, and verify answers. The author says two users can get very different results from the same model because their prompts carry different assumptions, and compares vague prompts to undefined behavior that produces plausible but wrong output. For practitioners, the useful lesson is operational: teams should turn prompting skill into reusable task specs, examples, acceptance checks, and review workflows before blaming ChatGPT or another LLM for every weak result.
The durable lesson is that prompting is closer to requirements writing than magic wording. A team that cannot define the task, constraints, evidence, and acceptance checks will often get weak AI output even from a strong model.
What happened
PlainEnglish published a July 7, 2026 essay arguing that AI is exposing a missing workplace skill: asking clearer questions. The author says public attention has moved from choosing the best AI tool, to finding the best prompt, to choosing a model subscription, while the deeper issue is the quality of thinking behind the request. The essay uses ChatGPT as the familiar cultural reference point and argues that two people can use the same model for a similar task and get very different results.
Technical context
The article's strongest analogy is to undefined behavior: a vague prompt may not fail loudly, but it can return a polished answer that is wrong in a way nobody notices until it is used. That is a useful framing for LLM workflows because many failures come from missing context, unclear success criteria, hidden assumptions, or no verification step after generation.
For practitioners
Teams can turn the essay into an operating habit by writing prompts like lightweight specs. Include the user goal, constraints, examples, unacceptable outputs, data boundaries, and a check for what would prove the answer wrong. For repeated work, move those choices into templates, evals, review checklists, and product workflows so quality does not depend on one person's private prompting instincts.
What to watch
Watch for organizations that stop treating prompt skill as individual craft and start managing it as shared process. The higher-leverage work is not just better wording; it is reusable task definitions, testable outputs, and explicit review loops for model-generated work.
Key Points
- 1The essay's practical value is treating prompt quality as a specification problem, not a magic wording trick.
- 2Teams get more reliable LLM output when prompts include context, constraints, examples, and acceptance checks.
- 3For production workflows, individual prompting habits should become shared templates, lightweight tests, and review loops.
Scoring Rationale
The essay offers useful practitioner guidance about prompt clarity, task specification, and verification workflows. It does not introduce new research, benchmark data, or tooling, so the impact is solid but modest.
Sources
Public references used for this report.
Practice with real Ad Tech data
90 SQL & Python problems · 15 industry datasets
250 free problems · No credit card
See all Ad Tech problems