Your Value-Add Is Showing
A junior PM sent me a Notion doc last week. I searched for “value-add.” Forty-seven occurrences. She didn’t write one of them. She confirmed this when I asked.
The worst part: before AI, she would have written twelve of them anyway. The rest of us would have nodded along, maybe rolled our eyes privately, and moved on. The signal-to-noise ratio in corporate communication has always been terrible. We just had plausible deniability — we assumed the speaker chose those words for a reason.
AI removed that assumption. And it turns out the assumption was doing all the work.
Jargon Was Never a Shortcut
The standard take on corporate jargon is that it is lazy — people use it because they have not done the thinking. This is comforting but wrong. Jargon was hard work. You had to learn the codes: “touch base” means meeting, “bandwidth” means availability, “circle back” means no, “alignment” means we fought about it and I lost.
Learning this vocabulary took years. Using it correctly required social calibration. People did not use jargon because they were lazy. They used it because not using it made them sound naive. The cost of plain speech in a corporate environment was sounding like you did not belong there.
Jargon was a signaling currency. It proved you had been initiated. It proved you understood how the game was played. And most importantly — it let you say things that could not be held against you later. “We need to align on strategic priorities” cannot be quoted in a post-mortem. “We should cut the marketing budget” can.
This is the function that mattered. Jargon was strategic ambiguity dressed up as professional polish. It let you sound decisive while committing to nothing. It let you disagree without disagreeing. It let you attend thirty meetings a week and never once say what you actually thought.
What AI Actually Revealed
We trained language models on annual reports, mission statements, all-hands decks, strategy memos, and LinkedIn thought leadership. The most carefully curated, least honest writing that exists. Every sentence engineered to sound certain while committing to nothing. Every paragraph optimized for maximum scanability and minimum accountability.
The models learned perfectly. Of course they did. This is the easiest pattern in the world to reproduce — form with no content, authority with no claim, alignment with no position.
And now we are mad at the output.
But the output is just a mirror. The outrage about AI-generated sludge is directed at the wrong target. The training data was us. Every quarterly review that said “we are well-positioned to leverage our core strengths” instead of admitting the product was not selling. Every email that said “let me circle back on that” instead of “I do not know the answer.” Every strategy deck that used “ecosystem,” “paradigm,” “synergy” to fill the space where a specific plan should have been.
We trained AI to write like we write. Now we are shocked that no one wants to read it.
The Complexity Backlash
The same dynamic plays out in solutions. A consultant shows up with a 14-step framework, seven integration points, a custom model, and a dashboard with seventeen panels. The problem: they need a spreadsheet and a weekly email. But a spreadsheet sounds like they are not doing enough. A framework sounds like they are thinking.
This is solution-space jargon. It is the same behavior — using impressive form to obscure the absence of direct experience. The stakeholders who roll their eyes at “synergy” are the same people who roll their eyes at “our proprietary multi-agent architecture.” The tell is the same: when the complexity of the description exceeds the complexity of the problem, someone is performing instead of working.
The backlash against “obsessive complex solutions from AI” is not really about AI. It is about a newly calibrated bullshit detector. People have seen enough AI output to recognize the shape of generation: the right structure, the confident tone, the complete absence of friction. They cannot articulate it, but they feel it. And they trust it less.
What Cannot Be Faked
There is exactly one thing a language model cannot do well: report a specific, flawed, first-person experience.
Try asking an LLM: “What happened when you deployed that change on a Friday afternoon?” It will give you a generic paragraph about best practices. It cannot tell you about the time you pushed to prod at 4:55 PM and spent the weekend reverting a migration. Because it was not there.
This is the only remaining signal. The sentence “we tried this thing and it broke” is now more credible than “our data-driven methodology ensures optimal outcomes.” Not because the former is more honest — but because it is harder to generate.
The premium has inverted. It used to be on sounding like you had it all figured out. It is now on sounding like you have actually been in the room. Specificity cannot be faked at scale. Flawed first-person accounts cannot be generated from a corpus of annual reports. They have to come from experience.
The Uncomfortable Part
We might not actually want to stop.
Jargon was serving a real purpose. It let us navigate organizations without making enemies. It let us attend meetings where nothing was decided without admitting that nothing was decided. It let us write strategy documents that sounded important enough to justify our salaries without committing to anything measurable enough to get us fired.
Strategic ambiguity is not a bug in corporate communication. It is the feature. And AI did not destroy it. AI just charged us rent on it — the rent being that everyone now knows the words are empty.
The question is not whether we can learn to communicate better. The question is whether we can function in organizations that reward clarity and punish the person who provides it.
AI made jargon honest by showing everyone the receipt. If we still cannot bring ourselves to write “I do not know” instead of “let me circle back on that,” the problem never was the vocabulary.