There was a time when the corporate world had its own unique — but entirely meaningless — language. We were moving the needle, boiling the ocean, constantly taking things offline, circling back, doing blue-sky thinking, and peeling the onion.
Corporate jargon was irritating, but at least I (these days people refer to me as a “veteran” journalist — a reference to my age, of course) knew what it meant. Or pretended to. But that seems a distant memory in the AI age.
Today, a visitor wandering into a typical corporate meeting could be forgiven for thinking they had accidentally stumbled into a machine-learning laboratory. “We need to prompt the team better.” “Let’s not hallucinate numbers in the presentation.” “That idea has great signal.” Et cetera.
What was once the language of engineers and data scientists has quietly escaped the server room and occupied the boardroom.
Take “prompting”. In AI, it refers to instructing a model. In offices, it now means anything from managing a subordinate to persuading a boss. A manager no longer briefs her team; she “prompts” them. Employees no longer need guidance; they need “better prompts”.
Then comes “hallucination” — perhaps the most dramatic migration of all. For decades, employees merely made mistakes. Today they hallucinate. An incorrect sales figure? Hallucination. An optimistic forecast? Hallucination. An executive claiming he always supported a failed project? Advanced hallucination.
A manager’s inbox now reads like dispatches from a parallel dimension. One colleague wrote: “Can you review this deck? I’ve been iterating on the prompts but the outputs still feel a bit low-temperature.”
A CEO in his mid-50s recalls a vendor presentation that opened with: “Our solution uses retrieval-augmented generation with a fine-tuned model, so you get the accuracy of your proprietary data without the risk of the model going off-distribution.” He sat through the whole thing before asking if it came with a user manual. For him.
The spread of AI vocabulary has been astonishingly swift because, unlike traditional corporate jargon, it sounds vaguely intelligent. Saying a team is “overfitting” to historical trends sounds considerably more sophisticated than saying it is stuck in the past.
Describing a colleague as having “high signal and low noise” feels considerably kinder than calling everyone else in the meeting annoying. And guardrails have already replaced governance, which itself replaced oversight — the endless corporate tradition of finding fresher words for the concept of “please stop before something goes badly wrong”.
Performance reviews are particularly fertile ground. Employees are advised to improve their “reasoning capabilities”, reduce “latency” in decision-making, and increase “throughput”. A slow worker is no longer slow; he has latency issues. Someone who sends too many emails generates excessive noise.
The star performer has become a “10x human” — apparently one evolutionary step short of becoming software. Low-temperature colleagues, meanwhile, are precise, conservative, and reliable, which is corporate AI-speak for “not much fun at parties”.
Promotions could soon be described as “algorithmic upgrades”. Turf wars may become “model competitions”. The annual budget round could be renamed “compute allocation”.
In any previous era of corporate life, telling someone that your system was hallucinating would have triggered an HR investigation and possibly a wellness check. Today, it is said freely, even fondly, in quarterly reviews. “The model hallucinated a bit on the client data, but we caught it.” Everyone nods. No one calls a doctor.
Somewhere right now, a 26-year-old is telling his manager that the team’s “loss function needs recalibrating”. The manager is nodding. He has no idea what any of it means. But then, he never really did. He just used to be better at hiding it.
The annual office party will probably survive — perhaps rebranded, at some point, as a “large-scale cross-functional human networking event”. But make no mistake — the essentials will remain unchanged. People will still gather around the food, make the same resolutions they made last year, and leave a little earlier than they promised.
Some technologies change the language. Very few change the species.
