工作中如何与同事沟通,减少返工他人质量不佳的 AI 内容
Tired of spending time fixing other people's AI slop at work? How to talk to them about it
职场教练 Melody Wilding建议,面对同事或上级交来的疑似 AI 生成、仍需大量修改的材料,应围绕共同目标沟通,并指出具体问题,避免直接指责对方使用 AI。她建议说明核验和编辑耗时,协商让对方先检查草稿、提前标注 AI 使用情况,或调整 AI 参与工作的方式。
You open slides you got from a teammate for a joint project presentation you're giving. Within minutes you suspect that it's largely AI-generated. The stats are incorrect and the writing is littered with generic phrases that sound plausible at a glance but, upon inspection, say nothing of substance.
AI was supposed to make work faster and easier, and sometimes it does. Eighty-five percent of surveyed workers, all of whom were active AI users, said the technology saved them between one and seven hours per week, according to a Workday study of 3,200 full-time employees published in January.
But as the author of "Managing Up" and an executive coach to high performers at companies like Google, Amazon and JPMorgan, I've seen a new problem emerge. You may regularly find yourself on the receiving end of what looks like finished work that still needs hours of editing. The Workday survey found that nearly 40% of the time "saved" from using AI was then lost to "rework" like verifying, correcting and rewriting.
Calling that extra work out to your colleagues and bosses can feel awkward. You might use AI too; 40.5% of employed U.S. adults ages 18 to 64 reported using AI for work in Q3 2026, according to data from the Real-Time Population Survey. So you don't want to sound hypocritical. Plus, you can't always be 100% sure someone used AI, even if it seems obvious to you.
So how do you address the extra work landing on your plate without the conversation becoming confrontational?
Start with a shared goal
Begin by showing that you're not trying to embarrass anyone or prove them wrong. You're on the same side.
Frame your feedback in terms of working toward an outcome that preserves the quality of the work and the team's reputation. That way, your intentions are clear, which makes the discussion less adversarial from the outset.
Try this:
- "We both want the executive team to be happy with this."
- "Since both of our names are on this, I hope you'll agree it needs to hold up to scrutiny from the client."
- "The goal is for the final version to have the most accurate and up-to-date numbers."
Narrow down your concerns
Avoid saying, "this whole thing looks AI-generated," especially if you can't prove that. A blanket accusation takes the focus off of what needs to change and puts the other person on the defensive.
Instead, point to concrete issues such as vague wording, missing context, hallucinated examples or off-brand visuals.
Try this:
- "A few things jumped out to me that we need to fix before this goes to leadership."
- "There are two sections where the data could better match what we had previously discussed."
- "The formatting is different throughout, so I think that needs another quick pass."
Mention AI diplomatically
This is particularly important when speaking with your boss or someone senior, since you don't want to suggest that you're judging their choices.
It can help to acknowledge if you use AI too, which makes it clear that the extra work and the impression the report or presentation could make are the problems, not the tool itself.
Try this:
- "I could be wrong, but some of this reads like AI was involved. I use it too, so that's not the issue. I'm more concerned about making sure we have enough time to edit it before the meeting."
- "Some of the phrasing and design choices in a couple sections have an AI-generated feel that people are getting more sensitive to. Since I also use ChatGPT and Claude in my work, I'm pretty familiar with some of the 'tells,' and I'd rather remove those so no one makes false assumptions."
- "People are getting quicker to label certain things AI-generated even when they're not and I'd hate for that to distract from the quality of the work."
Set expectations for next time
Be specific about how much time it's already required or will involve to clean up the work and whether it's taken away from other priorities. The other person may not realize the cost unless you explicitly say it.
Depending on the relationship, you might ask them to review AI-generated content before sending it over, limit when AI is used or simply flag it up front so you can plan accordingly.
Try this:
- "I've spent time redoing this, which meant pushing back a project for my boss. I hope you understand that's not sustainable. So going forward, can you give AI drafts a quick glance before sending them my way? That'll help me focus my time on core changes."
- "One thing that would help is knowing when AI was used. That gives me a better sense of how much verification and editing may be needed before it's ready to go so that we don't miss any deadlines."
- "Because the nuance that comes from your personal experience matters so much for this assignment, what do you think about the initial thinking coming from you? Totally fine if it's messy, and we can always use AI to clean it up later."
Remember, your colleagues and leaders are trying to work faster and keep up with ever-growing demands, just like you are. When you push back without shaming them, you show that you respect their intentions while still protecting your time.
Melody Wilding, LMSW is an executive coach, human behavior professor and author of "Managing Up: How to Get What You Need from the People in Charge." Get her free training, 5 Steps to Speak Like a Senior Leader, here.
来源:CNBC · cnbc.com