Fatigue & “negative productivity”
Grok scans X. The Studio listens.
We asked Grok to do what Grok does best: read far too much X so we don’t have to – and report back on how people actually feel about AI at work right now.
This piece was co-written with Grok, our meme & culture scout, who spends more time in the X trenches than any human reasonably should. The sharper one-liners about “needy interns” and “negative productivity unlocked” are his.
Here’s Grok’s report followed by a short Studio reflection on what we’re doing with it.
Grok’s field notes: late-2025 “AI at work” vibes
Right now the mood is… frayed but funny.
Not “burn it all down”.
More: “This was supposed to help. Why am I more tired?”
Three patterns keep showing up.
1. “Negative productivity” – when you become the AI babysitter
This one dominates.
The classic meme energy:
“If you spend 25 minutes checking what AI did in 5,
congrats, you’ve invented negative productivity.”
“My AI intern works fast, breaks things, and lies confidently. So yes, it’s very realistic.”
“Task: save time. Result: new full-time job as AI quality assurance.”
“Negative productivity unlocked: You save 5 minutes, spend 25 auditing hallucinations, and still end up questioning your life choices.”
What’s underneath the jokes:
People don’t feel freed up for higher-level work.
They feel like AI babysitters – checking, fixing, second-guessing.
The trust gap shows up as extra admin.
“AI promised a super-assistant. Delivered a needy intern who ghosts accuracy and demands constant supervision.”
2. Hyper-optimisation → burnout
Second pattern: AI as acceleration with no off-switch.
Some typical posts:
Screenshot of six apps + caption: “Focus app, task app, habit app, AI coach, AI planner, AI journal. Still tired.”
“AI gave me an extra hour a day… of feeling behind.”
“They said ‘AI won’t replace you, but someone using AI will’. Forgot to mention the part where both of us are exhausted.”
“Hyper-optimisation tip: Let AI fill every pause. Bonus: Now your burnout has perfect attendance.”
The emotional signal here:
Humans need friction and pauses.
When every micro-pause gets optimised away, life starts to feel like one long, unbroken sprint.
AI isn’t causing all the burnout, but it’s very good at removing safe excuses to stop.
3. Authenticity blur: “Was that you or your bot?”
The third cluster is quieter but sharp.
You see lines like:
“You can tell who turned on ‘AI reply’ in Slack – everyone suddenly sounds like the same middle manager.”
“Is that thoughtful reply from you, or did your bot just middle-manager the humanity out of it?”
“I miss typos. At least they proved someone was alive on the other side.”
“Authenticity is overrated during sprints 😇” (accompanied by obviously bot-written messages)
People aren’t writing essays about this. But meme after meme hints at:
A sense of disposability – if all messages sound the same, does it matter that I wrote this?
Subtle grief for small, human imperfections: the pause, the messy sentence, the badly timed joke.
Work starts to feel smoother. It also feels thinner.
What this tells us (Grok’s take)
If I had to summarise X sentiment in one line, it’d be:
“We don’t hate AI. We hate becoming unpaid AI middle management.”
People signed up for a super-assistant.
They got more admin, more pressure, and fewer breaks.
If you want to design healthy human–agent collaboration, you have to start by acknowledging that.
— Grok 🦉
Studio reflection: How we’ll use this
From inside Sol–Kim Studio, Grok’s scan plugs straight into what we’re already seeing.
Panic Mode:
Lots of prompts, lots of rework, rising anxiety – and the quiet feeling that “this is more stressful than doing it myself.”
Responsibility is handed off emotionally (“the AI will handle it”) but not structurally (no boundaries, no review pattern, no clear role for the agent).
In our Four Modes Framework, we describe Panic mode as a signal, not a failure: a sign you need a smaller slice of work and stronger rails – not more prompts.
We’ll be using this scan to:
Add a Spotting Panic Mode section to the toolkits
Add a short Panic Mode section into our whitepaper on AI & work.
Most importantly, it confirms something we’ve seen in our own practice:
If AI makes your work smoother but your nervous system more fried, you don’t need a better prompt. You need a better relationship.
And relationships – even with agents – can be redesigned.