The honeymoon phase is over. For two years, we were told AI would unlock our full potential. Now, the CEOs who pushed that narrative are scrambling to clean up the mess. The latest to admit it? Shopify's Tobias Lütke.
He's coined a term for the problem: "slop grenades." You know the type. That email from a colleague that's 800 words long, perfectly formatted, and says absolutely nothing. The code that's 50 lines longer than it needs to be. The report that looks great but has broken links and zero insight.
It's AI doing the work, but nobody taking responsibility. And it's making everyone's job harder.
The 'Workslop' Epidemic
Lütke didn't mince words on The Knowledge Project podcast. He described a workplace where employees are "tossing" these unexamined AI outputs at each other. The result? More headaches for everyone on the receiving end.
"We call those 'slop grenades' that people toss at each other," he said. "That's definitely a bad thing."
He's right. The problem isn't AI itself. It's the lazy application of it. Instead of using AI to synthesize a point, people are letting it generate a "big missive" that wastes everyone's time. You don't read it. Your colleagues don't read it. But they have to review it anyway.
This isn't just anecdotal. Researchers at BetterUp Labs and Stanford's Social Media Lab surveyed 962 workers. Over half (52.7%) admitted to sending workslop. Over a third (38%) said they receive it, costing them an average of 3.4 hours per month just to fix it.
That's a full half-day of work, gone. Every month. Just cleaning up after someone else's AI.
Why 'Slop' Is Worse Than Human Error
Here's the kicker: workslop is more dangerous than regular bad work. When a human makes a mistake, it's usually obvious. You can see the flaw, the gap in logic, the missing piece.
AI slop looks legitimate. It's well-structured. It uses the right buzzwords. It looks like a finished product. But it's hollow. It lacks the context, the nuance, and the actual utility that makes work useful.
The survey found that receiving workslop actually damages relationships. People view colleagues who send it as less competent and less friendly. Over a third of respondents said they actively want to avoid working with these people.
So, sending slop doesn't just waste time. It makes you look bad. It erodes trust. It's a career killer.
The Duolingo Backtrack
Lütke isn't alone in his reversal. Duolingo's CEO, Luis Von Ahn, was the poster child for "AI-first" companies. He evaluated employees on AI usage and replaced contractors with bots. But even he admitted he got carried away.
"We may need to write 1,000 different stories for people to learn a language, then you'll find that 20% of the things were just pure slop," he told Fast Company.
When you scale AI, the slop scales with it. And it gets harder to catch.
The Minimalist Fix: Use AI for Precision, Not Volume
So what's the answer? Do we throw out AI entirely? No. That's throwing the baby out with the bathwater. The answer is to be intentional.
The problem isn't AI. It's using AI for everything. It's using a sledgehammer to crack a nut.
Here's my rule: Use AI for tasks where precision matters more than volume.
Don't use AI to write a 2,000-word essay on a topic you know nothing about. Use it to draft a specific, data-driven email that you then edit. Don't use it to generate 50 variations of a logo. Use it to refine one concept.
This is especially true for financial tasks. You don't want AI slop in your invoices. You don't want a PDF with a broken link or a miscalculated total. That's not just annoying—it's unprofessional. It costs you money.
That's why tools like Invoice Gini are built differently. Instead of letting AI "go nuts," it's designed for a specific, high-stakes task: invoicing. You say what you need, and it generates a professional, accurate PDF. No fluff. No broken links. No slop. Just a clean, finished product that you can send to a client without embarrassment.
It's AI with a purpose. AI that respects your time and your client's time.
Stop Tossing Grenades
The lesson from Shopify and Duolingo is clear: AI is a tool, not a replacement for thinking. If you're using it to generate unexamined work, you're not being productive. You're being lazy.
And it's costing everyone around you.
Next time you're about to send that AI-generated email, read it. Actually read it. If it doesn't add value, delete it. If it's too long, cut it. If it's wrong, fix it.
Your colleagues will thank you. Your clients will thank you. And you'll actually get more done.
Source: Shopify CEO says employees' 'slop grenades' are making more work for everyone...