How to get coworkers to stop giving me ChatGPT-generated suggestions?
Workers are increasingly frustrated by colleagues who paste unvetted ChatGPT output into technical workflows, creating extra review labor and undermining trust in subject-matter experts. Commenters argue that the core problem is not the tool itself but people treating AI as an infallible authority, despite its tendency to generate plausible but incorrect answers. Suggested responses range from educating teams on AI’s limits and setting clear process guidelines to escalating persistent issues to management when they start to disrupt productivity.
Nature of the problem
- Non-technical coworkers (artists) use ChatGPT to generate technical suggestions they don’t understand.
- The developer must spend time validating and then explaining why many of these ideas are wrong, to people who lack the background to follow the explanation.
- This resembles older patterns like “I read on the internet that…” or copy-pasted StackOverflow/Google answers.
Using ChatGPT against itself
- Several suggest pasting the bad suggestion back into ChatGPT and asking it why the solution doesn’t work, then forwarding that explanation.
- Examples show ChatGPT confidently giving wrong code, then immediately apologizing and reversing its answer when asked “Does that work?”.
- Some see this as an efficient way to debunk; others view it as cult-like reliance on the tool.
Quality, responsibility, and critical thinking
- One camp: what matters is whether suggestions are good, not whether they’re AI-generated.
- Counterpoint: source does matter for nuanced work; ChatGPT often invents plausible but fake solutions.
- People stress that anyone forwarding AI output should be able to explain and defend it, and that using ChatGPT without verification is disrespectfully offloading QA onto others.
- Analogies: power tools needing skilled operators, non-malicious “Gish gallop,” Brandolini’s law, and junior devs merging untested code.
Role of management and team dynamics
- Disagreement over whether to “tattle to the boss”:
- Some say managers are explicitly paid to fix cross-team friction and protect IC time.
- Others argue adults should first set boundaries directly (“please stop sending ChatGPT suggestions”).
- Underlying issue is seen as lack of trust in the developer’s expertise and dysfunctional planning/feedback processes.
Comparisons to other tools and to art AIs
- Parallels drawn to past tech shifts: Photoshop, cameras, web search, StackOverflow.
- Some argue AI mostly raises the floor for weaker practitioners; others say that’s true of any tool.
- Artists using Stable Diffusion are invoked both as analogy and as a possible way to highlight their own replaceability if they overvalue AI.
Communication and AI fatigue
- Concern that AI-written emails and documents (e.g., via Copilot) will bloat communication and reduce willingness to read, especially when people turn short messages into PR-style walls of text.
- Some note you can also use LLMs to compress such fluff, but average users and bosses often equate length with quality.
Broader attitudes toward LLMs
- Strong skepticism: claims AI hype is driven by marketing; experts often find outputs unreliable or nonsensical.
- Others acknowledge serious flaws but still see LLMs as useful assistants when used critically and with proper validation.