There’s a viral phrase going around HR circles this year: “workslop.” It describes the specific hell of a coworker who sends you a suspiciously polished, oddly hollow email, which you then run through your own AI to decipher, before sending back a reply that they will, in turn, feed to their AI. One employee recently told a reporter she was fairly sure her boss’s AI was talking to her AI, and that she’d been reduced to a courier between two chatbots. Nearly two-thirds of professionals now say they spend six-plus hours a week cleaning up this stuff. We built machines to remove friction from our jobs, and instead built an elaborate telephone game where nobody has to feel anything, including being responsible for their own sentences.
What “workslop” is describing isn’t really new. Ross Blankenship and I spent a year researching this exact pattern for our book, Friction: How Tension, Emotion, and Change Reveal Better Leaders, long before anyone had a name for it. We’ve been finding ways to avoid dealing with feelings at work for a long time, we just used to say, “I don’t do feelings” or “that’s not my department” or “sounds like an HR thing” instead of “let me ask ChatGPT.”
Emotions are a Signal.
One core argument of our book runs a little against the grain of most workplace advice these days: friction, that knot in your stomach before a hard conversation, the irritation when someone talks over you, the dread that shows up around Sunday at 6pm, is important information. When Ross and I surveyed nearly 1,200 U.S. workers for our National Emotions Study, 56% told us they believe leaders feel more pressure to hide their emotions than everyone else does, and over half said that when leaders do hide their emotions, it creates more problems than it solves.
Suppressing emotion is, by multiple measures, expensive labor, it’s associated with worse memory, higher physiological stress, and weaker social connection. The person “keeping it together” in the meeting is doing something exhausting.
And that cost doesn’t stay contained to the person doing the suppressing. When a leader is emotionally closed, such as flat affect in hard conversations, deflection the moment things get personal, or a performance of certainty they don’t actually feel, the team reads that signal and adjusts. People learn, fast and without being told, that openness is risky here, and they start suppressing too. What follows is lower information flow, slower error correction, less creativity, and higher turnover among exactly the people you most want to keep. This is where emotional intelligence (EQ) actually earns its keep: not by eliminating friction, which isn’t possible or even desirable, but by determining whether that friction gets metabolized into better decisions or calcifies into exactly the costs above. EQ is the skill set that decides which way it goes, and the flip side of that cost is just as measurable. Organizations that invest in building EQ see, on average, 19% higher revenue growth and 63% lower turnover. The hidden cost of emotional friction isn’t hidden because it’s small. It’s hidden because it shows up on a spreadsheet as retention, error rates, or time-to-resolution, not “emotions.”
And some generations are paying that cost more than others. Gen Z workers in our survey reported the highest emotional drain of any cohort, hitting burnout roughly 17 years earlier, on average, than prior generations. Not because they’re softer, but because they were raised on the language of emotional openness and then handed jobs built by people who were raised on the opposite. That gap, what we call a “norms-conditions mismatch,” is friction with nowhere to go.
The Amodei Problem
Which brings us back to the robots. In his essay “Machines of Loving Grace,” Anthropic CEO Dario Amodei sketches an optimistic future where AI’s usefulness runs into hard limits: it’s extraordinary at pattern-matching within known parameters, he writes, but bad at knowing when the parameters have changed, or at weighing values an optimization function can’t capture. That, he argues, stays a human job, along with the harder-to-automate work of judgment under ambiguity and relational attunement. He goes further and suggests that as AI absorbs more of the technical floor, meaning increasingly comes from human relationships and connection rather than economic output.
Ross and I arrived, from a completely different angle, at almost the identical conclusion. Sixty percent of the workers we surveyed already sense this. They believe emotional intelligence will only matter more as AI takes over routine cognitive work.
But here’s the catch we kept running into while writing the book, one Amodei’s essay doesn’t quite reckon with: those same capabilities he’s counting on humans to protect — judgment, attunement, EQ — are exactly the muscles that atrophy first when you offload the friction of building them.
This isn’t just a hunch. Researchers studying students who use AI to write essays have found measurably lower cognitive engagement in the task, along with weaker recall and investment in the finished product, than students who wrote without it. The researchers call the pattern “cognitive debt”: shortcuts today that quietly compound into deficits tomorrow. There’s no reason to think emotional and relational skills work any differently. Judgment under ambiguity gets built by sitting in ambiguity, not by asking a model to resolve it for you. Attunement gets built by staying in an uncomfortable conversation long enough to actually read the other person, not by having a tool draft your way around it. Every time we let AI do the metabolizing, the very capacity Amodei is betting on, (humans getting better at the irreplaceable stuff) has one less rep to work with.
We felt this ourselves, honestly. We tried using AI a few times to help organize this very book, and each time the ideas came back competent, tidy, and somehow meaningless. The struggle was the point. The false starts and awkward sentences and not-quite-knowing-what-we-thought-yet was the thinking that needed to come before the writing.
So while Amodei is describing a world where AI clears space for deeper human judgment, a lot of workplaces in 2026 are using AI to clear space for less of it, outsourcing performance reviews, meeting notes, even the emotional labor of drafting a hard email, until, per a recent CNBC report on “AI weirdness at work,” some managers are now banning notetaker bots from meetings just to force people to actually listen to each other again. Amodei’s essay is a bet that humans will double down on the irreplaceable human stuff. What we saw while researching the book suggests plenty of workplaces are currently taking the opposite bet by default.
What Actually Works
This is about being deliberate with the friction you keep, not about rejecting AI. Here’s what the research, ours and others’, points to.
Leaders go first, especially if you’re skeptical. If you’re the leader rolling your eyes at all this, you don’t have to buy the whole emotional intelligence pitch to still be on the hook, because the cost is actually operational. Every time you shut down, deflect, or perform a certainty you don’t actually feel, you’re not staying neutral. You’re setting the terms for information flow, error correction, and who stays versus who quietly starts job-hunting. Your team learns what’s safe to feel from what you do in the hard moments, and getting it wrong lands on retention and revenue, not just morale.
Stop treating culture as a personal problem. This is the least comfortable finding in our whole dataset: individual EQ training can’t outrun a punishing culture. Two equally self-aware employees, one in a culture that treats emotion as data and one that treats it as weakness, will not have the same experience. If leadership is the one modeling suppression, sending everyone to a workshop won’t fix it, leadership has to change first.
Reintroduce friction on purpose. Before opening an AI tool for anything that requires actual judgment, write one paragraph of your own thinking first. Reinstate the 30-minute one-on-one you replaced with a shared doc. Add a 24-hour pause before finalizing decisions your team has started making on autopilot. None of these undo AI’s usefulness, they just protect the specific human capacities Amodei is betting we’ll need to keep building.
The line connecting a Gen Z employee burning out 17 years ahead of schedule and a Silicon Valley essay about machine-accelerated utopia is thinner than it looks. Both, in their own register, are asking the same question: now that so much can be made frictionless, what are we actually trying to protect?
My answer: the parts of work that were never supposed to be easy in the first place.
Sources
- Blankenship, Ross, and Sass, Maggie. Friction: How Tension, Emotion, and Change Reveal Better Leaders. San Diego: TalentSmartEQ, 2026.
- Sass, Maggie, and Blankenship, Ross. National Emotions Study: Emotional Intelligence and Workplace Emotion Among U.S. Adults. Proprietary survey data. San Diego: TalentSmartEQ, 2025.
- Amodei, Dario. “Machines of Loving Grace: How AI Could Transform the World for the Better.” October 2024. darioamodei.com/essay/machines-of-loving-grace.
- Gross, James J. Research on expressive suppression and emotion regulation, as cited in Blankenship and Sass, Friction.
- Goleman, Daniel. “What Makes a Leader?” Harvard Business Review 76, no. 6 (1998): 93–102.
- Nadella, Satya, interview by Mathias Döpfner. “MD Meets.” Axel Springer podcast, November 29, 2025.
- Eagle Hill Consulting and UKG generational burnout research, as cited in Blankenship and Sass, Friction.
- Landymore, Frank. “Workplaces Have Gotten So Bizarre That People Are Just Sending AI Slop Back and Forth at Each Other.” Futurism, July 7, 2026.
- “Gatekeeping bots, piles of slop: Welcome to the age of AI weirdness at work.” CNBC, August 20, 2026.
- Adaptavist. Survey on AI slop and knowledge worker sentiment, 2026, as reported by IT Pro.