More than half of senior managers find it harder to judge their team's work since AI arrived, and 78% have changed how they evaluate people because of it. A clear, well-organised document used to prove something besides its content, and that second thing is what has gone missing.
A clean document used to prove effort
Writing a clear update, a tailored proposal or a tidy summary took hours, and you could not do it well without understanding the subject. So a manager who read one got two things at once: the information, and evidence that the person had done the work. A muddled paragraph or a generic summary told the manager where to look closer.
AI now gives everyone the clean version in seconds, whether they understood the subject or not. The manager reads the same kind of sentences as before, but the second thing is gone. Last time we looked at this, the point was that AI makes creating work cheap while managing it stays costly. Judging work is the expensive half of managing, and it just got more expensive.
Hiring shows what a fix looks like
Employers on Freelancer.com get around 60 applications within minutes, so a tailored proposal was the quickest way to tell good workers from weak ones. Once one-click AI proposals arrived, employers' willingness to pay for that tailoring fell sharply. A manager reading eight polished updates on a Friday afternoon is in the same position as that employer, with the same loss of information.
The researchers who studied the site suggest two ways out for bigger jobs. One is a live conversation. The other is a paid trial of the first tenth of the work before committing to all of it. Both work for the same reason: they test something the person has to do in front of you.
What this looks like for a manager
Take a regional sales manager with six reps who each send a weekly summary. A year ago, the weak summaries gave away who was losing grip on an account. Now all six read well. She could keep the written summary and add ten minutes a week with each rep, where she picks one number from it and asks why it moved and what they will do next. That is an hour a week, and it shows her what six polished pages cannot: who understands the account.
The same applies to what managers now say they grade. Many are weighing how well staff write prompts and decide what to hand to AI. Those skills also show up on the page only through the output, so the way to judge them is to watch someone work for a few minutes.
It applies to reviews too. Of the leaders in the same research, 39% use AI to draft performance reviews. That makes a loop: AI-polished work, judged in an AI-drafted review, with no point where a person checked what the employee understood. A review is better written from what the manager saw directly, with AI used afterwards to tidy the wording.
Expect written updates to count for less, and the short conversation where someone explains their own work to count for more. For employees that is good news. When everyone's documents look the same, the person who can explain theirs in two minutes is the one who gets noticed.