Workslop is Killing Your Culture

Somewhere in your company, there are people who are drowning. The worst part is that they're your most competent employees, the ones who not only care about the quality of the work product, but also have the skills to improve it.

This isn't a post about overwork, because that's not the root cause. This is a post about their coworkers who have discovered they can generate a 40-page document in a few minutes and then ask someone to make sure it's any good.

Defining Workslop

Late last year, the Harvard Business Review introduced the term workslop. I immediately fell in love with it, because it's what I see frequently in my AI literacy and advisory. Workslop is output that has all the markers of competence: structure, headers, and lots of words that belong in the context. However, like much slop, it fails to hold up to scrutiny.

Now, this isn't a new behavior. Anyone who has spent significant time in the workforce knows people who just... don't provide good value. In fact, I've seen many a useless person make a career of hopping from company to company and bullshitting just long enough to jump ship to a new company or team.

But at least that bullshit took some effort. So to use an IT term, it was rate limited.

AI has removed the rate limiter. The floor on how much output a bullshit artist can produce is now effectively infinite.

Workslop Relocates Work

If you hang out on social media, especially on LinkedIn, you'll see daily posts bragging about how much productivity the AI boosters are gaining. But you also see reports of burnout, disengagement, and the evidence of lower quality products and services is so apparent that the word enshittification has entered the public psyche.

I'm pretty confident at this point that the reason is workslop. Here's what I'm encountering in the field:

• The creator of the workslop does the easy part: they prompt, they skim, maybe do some light formatting.
• This person gets rewarded because they look productive, they ship fast, feel smart, and have artifacts with their name on it.
• The slop flows through the system until it reaches someone competent who cares. This person then takes on the hard part: real thinking, reading closely, finding and fixing errors.

One thing I internalized over a career of software development is that reviewing work is harder than creating it. Also, AI output is even harder to review than the average bad work because it sounds so convincing.

"Can you review this?" used to be a request for a second opinion. Now, it's a request for a first one.

So, the reason we're simultaneously seeing people claiming about "10x'ing their productivity" but the needle in quality, revenue, and overall productivity has barely moved is because the competent people are buried in cleaning up the slop.

This is a Culture Problem

I frequently blast leaders for pushing AI into all the things, even when it's not appropriate. But look, non-technical leaders get sold on things all the time. Even technical leaders get fooled. That's sad, but forgivable. But my personal leadership philosophy is that a good leader shields their workers from as much crap as they can, so that those workers can be productive.

What isn't forgivable is deliberately building incentives and then acting confused about the results. Your token-usage dashboard only measures quantity, not quality. And whatever you measure, you'll get more of. Any time spent fixing workslop and not pushing tickets looks bad in the system you've created.

The fact that you don't understand second order effects makes you a bad leader, regardless of what tools your team is using. It's the same reason why I hate NPS scores in education: they are not a good proxy for outcomes.

Here's what happens when leaders don't put a stop to workslop, or worse, reward those producing the slop with bad ideas like "AI token usage boards".

Collaboration Dies First

Competent people learn fast who the worksloppers are. Then, they do what competent people tend to do. They start routing around them. Declining the meeting, "getting to it next week", and excluding them from workgroup discussions.

Nobody reports it, because "Joe sends me nothing but slop", isn't a thing they can put in a status update, especially when the leadership team is full steam ahead on AI adoption.

So, they roll their eyes, avoid Joe as much as possible, and roll up their sleeves and do his work when they must.

Then, the Burnout

Eventually, the burnout sets in. The competent people keep absorbing the extra workload, taking on the responsibility for the slop until they're "cooked" as the kids like to say, and eventually they just stop caring.

This usually results in one or both of these behaviors:

1. They stop fixing other people's slop, and just let it through.
2. They start producing their own slop, because that's rewarded and rigor is seen as "not buying into the company direction".

This is a mixture of surrender and malicious compliance. Because your best people are smart, and they know when rigor and care becomes a liability.

Next, the Brain Drain

Competent people have options. That's what makes them competent people. They don't file a complaint, they leave. And you're left with the people who couldn't get hired elsewhere, along with a pile of slop that nobody wants to engage with.

The bad economy is providing some cover for this situation, because many people are afraid, but when things improve, there is going to be a mass exodus.

Finally, the Revenue

This is also where it's going to impact revenue... eventually. Let me tell you a story from personal experience:

We were working through discovery with a potential partner, and there was a team member who was obviously outsourcing their brain and work product to AI. Even simple Slack messages would come back verbose, filled with emdashes and AI tells.

I told my wife, Shelby, "that person is going to cost them some business if a client ever sees her work". We moved on, and I checked in with a connection 3 months later. Sure enough, this workslop producer ended up in a situation with a client, and cost the company a high 6-figure deal.

This is why workslop is so dangerous. It's a slow burn. It shows up a few quarters later as customers and partnerships start abandoning ship whether it's from poor QA, making decisions based on hallucination numbers, security breaches, or any other of the multitude of examples I could pull out of the news.

One of my most scathing complaints about AI boosters is that if they truly are seeing 10x productivity, why don't they have more time for QA?

How to Make it Stop

Now I'm going to talk to individual contributors. If you're in this situation and your leadership isn't going to, you know, do their job, you need to stop being polite about workslop. Here's two questions you should ask before you accept any review requests or work product from a known slop artist:

1. What is this for? What outcome does it achieve? This is how you check to see if the work is performative (ooh, look at all the activity I have!) or something of real value. If it's the former, send it back.
2. Which parts are yours, and which parts do you need me on? This is a simple concept. If they really need your expertise, they don't need you to read 40 pages. Highlight the sections, tell me the parts you're unsure about. If they can't do that after producing something, regardless of whether it's code, a doc, or a presentation, then they didn't review their own work. And I'm going to be harsh. That is contempt for you, the product, and the organization, and you should treat it as such.

A common theme of workslop producers is that they're not asking for collaboration, they're asking you to do all their thinking for them.

And if someone can't answer those questions, you can send them away guilt free. They'll either find another sucker, or they'll let it drop. After all, what are they going to do? Go to their boss and complain that you won't do their work for them?

AI is just a Tool

While it's popular to be entirely pro-or-anti-AI, at the end of the day it's just a tool. The culture I like to cultivate with my clients is that wherever AI does fit into a process, a human brain needs to be engaged before and after the generation step.

We talk about culture in my leadership training sessions, and it's important for leaders to realize that culture is less about writing your values down for everyone to see and more about what you tolerate.

If you tolerate workslop, you're not "building an AI-first organization", you're building a team where people stop caring. When it rolls downhill into revenue, you shouldn't act surprised.