A reader of my paper on the epistemic hazard of AI delegation asked a fair question: what does an organisation gain by calling in someone like me? Here is my answer, with its limits stated.
The problem comes first. AI can produce work that looks finished without producing a person who understands it (The Machine Is Doing It). The risk sits between "looks good" and "has been understood". The people best placed to see that gap have done the work for years and know where it usually goes wrong.
An experienced practitioner can help at two scales, and each changes something different.
One to one (parthans.com)
This is coaching and AI enablement for one person: a manager, a founder, a mid-career professional. It changes that person's judgment. They bring a real task, make their own attempt first, compare it with what AI produces, and learn from the difference. They also practise the conversations they have been putting off.
The change goes wherever the person goes. It does not change how their team reviews work.
Team (AI Content Ops)
I co-founded AI Content Ops with Anant Sabane. It works with existing corporate teams, mainly documentation, L&D and content operations, in hands-on programmes adapted to the team's own tools and content.
What changes here is shared: a written standard for acceptable output, a review that tests judgment, and the same questions asked before work is handed to AI. A trained individual can do this for themselves. A team needs it agreed between people.
Which one fits
It depends on where the gap sits. If the problem is one person's confidence in their own judgment, start one to one. If three people on a team check AI output in three different ways, it is a team problem, and one person's training will not fix it.
I have no outcome figures for either format. I am describing what each is designed to change, and I would rather say that than invent a number.
Where does the gap sit in your organisation: in one person, or between people?
Sources: The Machine Is Doing It. More at AI Enablement and aicontentops.online.
