Somewhere in Chicago a 40-year-old project director spent her summer second-guessing every message that arrived in her inbox. She had been paired with an artificial intelligence specialist who decided to hand his communications over to a set of agents. Important emails vanished. Texts never reached him. Client deadlines slipped. Eventually the only way she could imagine closing the loop was to hunt down his public speaking schedule and show up in person. At that point she was no longer sure the person on the other end of the contract was even real. “I will take a messy human over a robot any day,” she told me. “How do I get through to a machine?”
The Strange New Normal Of AI At Work
That story is not an outlier. It is a snapshot of the awkward, sometimes absurd transition most of us are living through right now. Generative tools have moved from novelty to daily fixture, yet the social rules around them remain half-written. Roughly half of workers say they use these systems at least once a week, and a surprising number rely on them multiple times a day. The vast majority taught themselves. More than half of workplaces still lack any formal policy. The result is a quiet collision of expectations, suspicions, and half-finished experiments that leave people exhausted and oddly lonely.
I have watched colleagues treat every polished paragraph as potential evidence of cheating. I have watched managers quietly redirect questions to a chatbot and then wonder why their teams feel less loyal. I have watched talented people spend evenings cleaning up content that looked professional but collapsed under the slightest scrutiny. The technology is powerful. The human layer surrounding it is still improvising.
When Coworkers Start Suspecting Everyone
The first and most immediate shift is a low-grade paranoia that settles into everyday interactions. An HR professional in London spent careful hours researching and drafting an email only to receive a compliment followed immediately by the question: “Did you use a chatbot on this?” She felt the air leave the room. The praise had been converted into an accusation. Her hard-earned judgment had been reduced to the skill of writing a decent prompt.
That moment is becoming common. People who once traded feedback freely now hesitate. A copywriter in Chicago describes her job as endless triage of other people’s generated drafts. When she suggests a clearer phrasing, the response is no longer collaborative curiosity. It is defensive pushback, as if the machine’s authority outweighs hers. “Is that what you believe,” she asks herself, “or is that what the model told you?” The question hangs in the air and changes the temperature of every conversation.
Research has begun to quantify the cost. Workers who openly admit using these tools are often rated as lazier and less deserving of stretch assignments. The perception gap is real even when the work itself is solid. In my own experience the suspicion cuts both ways. I have caught myself scanning a colleague’s document for the telltale smoothness that sometimes signals heavy reliance on a model. Then I catch myself and feel a little ashamed. The tool is not the enemy. The absence of shared norms is.
Piles Of Workslop And The Quiet Resentment They Create
If suspicion is the emotional weather, workslop is the physical debris. The term has entered the vocabulary for a reason. It describes content that looks finished on the surface yet lacks substance, accuracy, or genuine thought. Nearly two in five desk workers report receiving material of this kind, and the extra hours required to repair it add up fast. One founder of a communications firm in Michigan told me she lost ten thousand dollars rewriting an entire campaign because every piece the contractor delivered carried the unmistakable gloss of unexamined generation. Relationships strained. Trust evaporated.
A marketer in the Midwest put the frustration more personally. She can no longer tell whether the person who sent the draft believes it is good or simply decided the appearance of progress was enough. Either explanation is demoralizing. Both slow her down. She finds herself delaying replies because she already knows the next exchange will require mental energy she does not have. The emotional tax is invisible on any productivity dashboard yet very real in the daily experience of work.
It is hard to tell whether they think it is actually satisfactory or whether they know it is not good and are sending it anyway. Both possibilities create problems of their own.
I have sat through meetings where the slide deck was beautifully formatted and almost empty of insight. The room sensed it. No one wanted to be the first to say the emperor had no clothes. Later, in quieter conversations, the resentment surfaced. People felt their time had been treated as disposable. That feeling lingers longer than any single bad deliverable.
Job Insecurity Makes Everything Tighter
Layer the economic climate on top of the social friction and the tension rises further. Roughly one in five workers already feel their role is less secure because of these systems. One in three say the technology has made them more pessimistic about the job market. In a low-hire, low-fire environment many people are clinging to positions while scanning the horizon for the next wave of automation. The pressure from leadership is usually the same: move faster, do more with less, treat the new tools as the obvious solution.
Yet the returns have been uneven. Organizations that cut headcount in favor of increased spending on these systems have often failed to see the expected gains. Large-scale studies continue to show that the majority of generative investments produce little measurable return. The gap between executive enthusiasm and frontline experience creates its own strain. Workers are told to embrace the technology while simultaneously worrying that mastery of it may accelerate their own replacement.
One executive I spoke with admitted he had to call out his own team for producing documents that were articulate to the point of emptiness. The models can make anyone sound like the most polished communicator in the room. Closer inspection often reveals filler dressed up as strategy. The time other people then spend decoding that filler is never counted in the productivity narrative.
Social Offloading And The Thinning Of Human Connection
Perhaps the most subtle damage is the growing habit of treating the technology as a middleman between colleagues. About a quarter of users have redirected a coworker’s question to a model before answering it themselves. The stated reasons are reasonable on the surface: encourage self-sufficiency, protect focus time, reduce interruptions. The lived experience on the receiving end is often different. It can feel lazy, almost insulting. You came looking for a particular person’s judgment and received a generic search result instead.
When leaders do this with developmental conversations the effects compound. Research has linked the substitution of machine-mediated advice for genuine mentorship to higher desire to quit, elevated burnout, and weaker team coordination. There is a psychological cost when every interaction is filtered. Human connection is not the opposite of productivity. It is often the condition that makes sustained productivity possible.
I have watched talented junior people grow quieter after repeated redirects. They stop bringing the messy, half-formed questions that used to spark real learning. The surface efficiency looks good. The longer-term erosion of judgment and belonging is harder to measure and easier to ignore until it becomes expensive.
Growing Pains And Unintended Consequences
New tools always arrive with side effects no one planned. An architecture firm in Washington discovered that an AI-assisted notetaker had been automatically added to every meeting. Transcripts of sensitive discussions were distributed to the entire invite list. Context disappeared. Critiques that should have stayed private suddenly lived in permanent records. The chief people officer found herself managing a series of human-resources fire drills. The firm eventually set a simple rule: no automatic notetakers unless every participant agrees in advance. The goal was not to ban the technology. It was to keep human interactions human.
Surveys show employees are evenly split on whether workplace culture has improved or worsened since these systems became common. That split itself is telling. The same tool can feel liberating to one person and dehumanizing to another depending on how it is introduced and governed. The friction is less about the capability of the models and more about the absence of shared expectations.
In my view the most useful frame is to treat the current moment as an extended period of social improvisation. The technology is not introducing entirely new problems. It is putting a new spin on old ones: unclear accountability, uneven effort, poor judgment, and the eternal difficulty of balancing speed with care. Work was already complicated. Now an additional layer has been dropped into the mix and everyone is still learning the steps.
What Actually Helps Right Now
There are practical moves that reduce the weirdness without requiring a company-wide transformation. Clear norms around disclosure help. When someone is transparent about heavy reliance on a model, the suspicion drops. When teams agree that certain kinds of work still require human first drafts, the volume of workslop declines. When leaders refuse to outsource developmental conversations, psychological safety improves.
Simple process changes matter more than most people expect. Requiring a short human summary of any generated content forces the person submitting it to engage with the material. Setting explicit expectations that feedback will be treated as collaboration rather than accusation restores some of the earlier ease. Protecting a few meetings from automatic transcription preserves space for candid talk.
- Agree as a team on when disclosure of tool use is expected
- Require a brief human-authored summary with any heavily generated deliverable
- Protect developmental and sensitive conversations from automatic mediation
- Treat polished but empty work as a process failure rather than a personal one
- Measure the time spent cleaning up workslop so the hidden cost becomes visible
None of these steps are revolutionary. They simply reassert that the people in the room still matter more than the systems they use. In my experience the teams that recover their footing fastest are the ones that talk openly about the friction instead of pretending it does not exist.
The Longer View
The technology will keep improving. The social questions will remain. How do we preserve credit for genuine skill? How do we keep mentorship from being replaced by generic advice? How do we maintain the messy, inefficient conversations that sometimes produce the best ideas? Those questions cannot be answered by better models. They require deliberate choices about the kind of workplaces we want to inhabit.
I keep returning to the project director in Chicago. Her willingness to track down a speaking engagement rather than accept permanent radio silence feels almost old-fashioned in the best sense. She insisted on a human connection even when the systems made that connection difficult. That insistence is still available to the rest of us. The bots can filter the inbox. They cannot decide whether we will keep treating one another as people worth the extra effort.
The age of AI weirdness is not temporary. It is the environment in which the next decade of work will unfold. The organizations that navigate it with the least damage will be those that treat the human side of the equation as seriously as the technical one. Everything else is just more polished noise.
The stories keep arriving. Another manager describes watching a junior employee paste a model’s response into a client email without reading it. Another designer talks about the quiet grief of watching colleagues treat her carefully considered feedback as optional because a system offered a different suggestion. These moments accumulate. They change the texture of ordinary days.
Perhaps the most useful stance is a kind of pragmatic curiosity. The tools are here. They will be used. The only real question is whether we will let them quietly rewrite the social contract of work or whether we will keep rewriting that contract ourselves, one awkward conversation at a time. I know which version I prefer. It is the messier one, the one that still leaves room for the project director who would rather drive across town than accept silence from a machine.
That preference is not nostalgia. It is a practical recognition that the parts of work that matter most—trust, judgment, the willingness to stay in the hard conversation—still require people who refuse to outsource everything that feels inconvenient. The bots can handle the rest. The rest is still ours.
Everyday Friction In Practice
Consider a typical Tuesday. A product manager asks a designer for a quick reaction to a new flow. Instead of a conversation she receives a neatly formatted list generated in seconds. The list is competent and strangely lifeless. The product manager feels dismissed. The designer feels efficient. Both walk away slightly less willing to initiate the next exchange. Multiply that pattern across a week and the collaborative muscle begins to atrophy.
Or take the performance-review season. Managers under pressure to complete forms turn to models for polished language. The resulting documents sound professional and say almost nothing distinctive about the actual person. Employees notice. They feel seen less clearly. The developmental conversation that should have followed never quite happens because the written artifact already looks finished. Another small erosion of the human layer.
These are not dramatic failures. They are the ordinary background radiation of the current moment. Because they lack spectacle they are easy to normalize. Yet the cumulative effect is a workplace that feels slightly more transactional and slightly less alive. People still show up. They still produce. They simply invest a little less of themselves because the return on that investment has become less certain.
Reclaiming Agency Without Rejecting The Tools
Rejecting the technology wholesale is neither realistic nor necessary. The more interesting path is selective reclamation. Decide which interactions are worth protecting. Decide which kinds of output still require a human first pass. Decide which questions should never be redirected. These decisions do not require executive sponsorship. They can begin inside a single team.
I have seen small groups institute a simple ritual: any generated draft is accompanied by a three-sentence note describing what the human contributor actually thought. The quality of the subsequent conversation improves immediately. The draft is no longer an anonymous artifact. It is a starting point for real exchange. The difference is modest and surprisingly powerful.
Another group banned automatic meeting notes for anything involving performance, conflict, or career development. The meetings became shorter and more candid. People spoke more carefully because they knew the record would be human-curated. The technology was still available for less sensitive work. The boundary itself restored a sense of intentionality.
These experiments are imperfect. They require ongoing adjustment. What matters is the underlying posture: the technology serves the relationships, not the other way around. When that order is reversed the weirdness intensifies. When it is maintained the tools become useful rather than corrosive.
The Cost Of Pretending Everything Is Fine
Many organizations still treat the social friction as a temporary adjustment period that will resolve on its own. That hope is misplaced. The friction is structural. It arises from the mismatch between the speed of capability growth and the slower pace of cultural adaptation. Ignoring it does not make it disappear. It simply allows the quiet resentments and eroded trust to compound.
Leaders who acknowledge the awkwardness openly tend to fare better. A simple statement—“We know these tools are creating new kinds of friction and we are still figuring out the norms”—gives permission for honest conversation. Without that permission people default to private grumbling and quiet workarounds. The official narrative remains optimistic while the lived experience diverges further each month.
I have sat in rooms where the gap between the two was almost comic. Executives celebrated adoption metrics. Individual contributors exchanged knowing glances. The metrics were real. So was the exhaustion. Both can be true at once. The organizations that close the gap are the ones willing to measure the second truth with the same seriousness as the first.
A Final Note On Messy Humans
The project director who was ready to drive across town to find her missing collaborator understood something essential. Efficiency is not the highest value. Reliability of human connection still matters more than the smoothness of any generated reply. In the long run the teams that remember this will outlast the ones that optimize it away.
We are living through a genuine transition. The tools are extraordinary. The people using them remain ordinary in the best sense: fallible, creative, occasionally stubborn, capable of both remarkable care and careless shortcuts. The weirdness of the current moment is simply the sound of those ordinary people trying to figure out how to stay human while the systems around them grow more capable by the week.
That figuring-out is still underway. It will not be completed by any single policy or training program. It will be completed in thousands of small choices—whether to disclose, whether to clean up the workslop before sending it, whether to answer the question yourself or hand it to a model, whether to protect the meeting from automatic transcription. Each choice either thickens or thins the human layer. Over time those choices become culture.
I prefer the thicker version. It is slower. It is less polished. It leaves room for the project director who would rather meet a messy human face to face than negotiate forever with a gatekeeping bot. That preference is not anti-technology. It is pro-human. And in the end that is the only orientation that keeps the work worth doing.
