In a 2025 BetterUp Labs and Stanford survey of 1,150 US desk workers, 53% admitted sending AI work with little substance at least some of the time. Half of the people who received that kind of work saw the sender as less capable. If AI makes your work polished but you’re not sure it still carries your judgment, this newsletter is your next move.
One of my favorite podcasts is The AI Daily Brief with Nathaniel Whittemore. This week he ran an episode called How to Deslopify Your AI, and it put words to something I think about every time I sit down to work.
AI can write almost anything now. Emails, memos, proposals, reports. Fast, clean, and confident.
What it can’t do is decide what you think.
I use AI every week. It helps me research and draft, including this newsletter. But the draft is never the finish line. My real work starts after it. I review what AI gives me, I check whether it’s high caliber, and I add my own taste and preferences until it says what I actually mean.
That last part is the part nobody else can do for me.
Here is the trap. A polished draft feels finished. The sentences flow. The structure is clean. It sounds professional. So it’s tempting to hit send.
But polish is not a position. A document can be well written and still not say what you recommend, why you believe it, or what you’re willing to stand behind.
AI can give you the words. Only you can give them a point of view.
My Tía Pika has a saying for this. No, her real name is not Pika. It’s Barbara. We’ve called her Pika all my life, and I have no idea why or where that nickname came from, but that’s a story for another day. LOL
Here’s what she used to tell me:
“No todo lo que brilla es oro.” (Not everything that glitters is gold.)
It fits a lot of AI work. A draft can shine and still hold very little you can actually use. A memo that reads well but never makes a call. A report that summarizes everything and recommends nothing.
What makes work valuable is what you add to it. AI doesn’t have that unless you bring it.
The goal isn’t to write everything yourself. The goal is to make sure every piece of work carries a position you chose and can stand behind.
1. AI work with little substance is common, and many respondents reported sending it. In 2025, BetterUp Labs and the Stanford Social Media Lab surveyed 1,150 US desk workers employed full time about what they called “workslop,” work produced with AI that “masquerades as good work, but lacks the substance to meaningfully advance a given task.” 40% had received it in the past month. In BetterUp’s summary of the findings, 53% admitted sending it at least some of the time. The research was published in Harvard Business Review in September 2025. This isn’t only a problem other people have. It’s an easy habit to slip into.
2. The real cost lands on the sender’s reputation. In the same BetterUp summary, people who received workslop said they saw the sender as less capable (50%), less reliable (49%), and less trustworthy (42%). 32% said they didn’t want to work with that person again. These are survey answers people gave about themselves. Still, the signal is worth noticing. People judge the person behind the work, not the tool.
3. Confidence in AI and confidence in yourself were linked to different levels of critical thinking. Microsoft Research surveyed 319 knowledge workers about how they think critically when using generative AI. Higher confidence in the AI was associated with less critical thinking. Higher confidence in their own ability was associated with more critical thinking. Again, this is self-reported and not scientific proof that AI causes anyone to think less. However, it is a useful insight. The more we trust the tool, the less we tend to question what it gives us.
Put those together and the problem isn’t that AI writes. The problem is letting the tool supply the position, the evidence, and the confidence all at once. When you stay clear about which parts are yours, AI becomes a sharper thinking partner. When you don’t, it fills the space with the most likely answer and your name goes on it.
Why we believe it: For most of our careers, polish was proof of effort. A clean, well organized document meant someone had put in the hours. If you’ve heard me speak or joined one of our webinars, you’ve heard us talk about the old equation for success:
Effort = success
Before AI, that equation mostly held. Writing well took real time, so good writing read as good thinking.
Reframe: Polish is no longer proof of thinking. Now AI can produce a clean, generic draft in seconds, and that equation breaks. Effort still matters. It just isn’t what sets your work apart anymore. The new equation is:
Impact = success
Impact is what happens when you take generic artificial intelligence output and turn it into authentic intelligence. You do that with your lived experience, your contextual knowledge, your taste, your judgment, and your zone of genius. In practice, impact looks like a recommendation that helps someone make a decision, the context that fits the situation, and a next step the reader can actually act on. The old playbook for success doesn’t work anymore. We need a new playbook for having more impact with AI.
What to do: Before you share work you created with AI, find the sentence that states your position. If you can’t find it, write it. And if you’re not sure yet, that’s fine. A provisional view counts. So does “I haven’t decided yet, help me compare the evidence.” You don’t need a firm opinion before you start. You need one before you send. If you want AI to sound more like you as well, my past newsletter The Portable Context File: Why a Better AI Model Still Won’t Sound Like You covers voice. This week is about something underneath voice: the judgment.
You don’t need a new app. Use the AI tool your company already approved. Describe your work in general terms and leave out client names, financials, and anything confidential. The AI challenges you. You make the final call.
Run this 30 minute test:
Prompt to steal:
I need to write a [TYPE OF WORK, such as a decision memo, team recommendation, or executive proposal] for [AUDIENCE]. My current position is [WHAT I RECOMMEND AND WHY, OR “I haven’t decided yet, help me compare the options”]. What is at stake is [STAKES]. What I know from experience that you don’t is [CONTEXT]. What I am still unsure about is [UNCERTAINTY]. Before you write anything, act as a thoughtful skeptic. Give me the three strongest arguments against my position, the questions my audience is most likely to ask, and anything important I may be missing. Do not agree with me just to be helpful. Wait for my answers before you draft.
Ask AI to sort its draft into three piles: your facts, its inferences, and claims that need checking.
AI is very good at filling gaps with details that sound right. A plausible number, a confident example, a source that seems real. Once it’s woven into a polished draft, it’s easy to mistake that detail for your own evidence.
So before you share anything that matters, give your AI tool this prompt: “Label every claim in this draft as (1) information I provided, (2) something you inferred, or (3) a claim that needs a source. List the sources you used.”
Then you, the person sharing the work, check the sources yourself. Open the links and confirm that they support the claims and numbers. Read the first two labels with the same care, since AI can mislabel claims and its inferences need your judgment too. AI can organize the checks. It cannot approve its own evidence for you.
1. Audit one AI draft you already shared. Pick an email, memo, or slide from last week. Highlight every fact in one color and every assumption in another. Notice which assumptions you would defend and which ones simply came with the draft.
2. Give your next draft one detail only you know. A constraint from your team, a lesson from a past project, a reason the obvious answer won’t work here. One real detail turns a generic draft into your work.
3. If you lead a team, ask for the position before the polish. When someone brings you work created with AI, start with “What do you recommend, and why?” This isn’t about policing AI use. It’s about making judgment the expected part of the job.
Keeping your point of view at the center of AI work is a skill you build with practice and good company. I have a special invitation just for you as a ¡Ay, Ay, Ay, AI! reader.
On Wednesday, November 4, Nikki and I kicked off our newest FlipWork Sprint™ cohort with a great group of senior leaders and executives, and I’m excited for what’s ahead with them. There is still time to join. We are taking late registrations through Wednesday, November 11. We meet once a week for 60 minutes, and you put additional time into the FW Sprint work between sessions.
Over the eight weeks, you’ll turn your expertise into systems powered by AI for real work. Those systems are things like Markdown files (plain text documents your AI tools can read), skills files (written instructions for how a piece of work should be done), documents, and workflows that capture your unique taste and your judgment logic. Your Reinvention Passport captures how you think and how you decide, with frameworks that help pull out your essence.
This is where you put your zone of genius to work, drawing on your expertise and the diversity of thought that only you bring to the table. AI helps carry the work. You stay the one who decides and reviews. You finish with a Capstone that shows the before and after, and earn the FlipWork Certified: Agentic Leader credential.
Most of our FW Sprints are closed cohorts for enterprise clients. Only about once a quarter do we open a public cohort that individuals can join on their own. Most participants are sponsored or reimbursed through their company’s professional development or education benefit.
Want to know what it is really like? Hear directly from past participants.
AI can write it. That part is settled.
The question is whether your work still says something. Whether someone reading it can tell what you recommend, why you believe it, and that you are willing to stand behind it.
Use AI for the words. Keep the judgment for yourself.
No todo lo que brilla es oro. Not everything that glitters is gold. Tía Pika was right. Polished words are easy now. Make sure yours carry something worth keeping.
¡Hasta la próxima, un abrazo fuerte! (Until next week, a big hug!)
P.S. ♻️ Know someone who uses AI every day and wants their work to still carry their own thinking? Send them this newsletter and invite them to subscribe to ¡Ay, Ay, Ay, AI!. If they are ready to build that skill with other senior leaders, share the FW Sprint invitation with them too.