Why I Didn't Use AI to Review 300 Resumes
Situation
I was hiring for a program manager role. About 100 resumes came in the first day. By the weekend there were over 300.
Recruiters see piles like this on every role, every week. I was seeing it once.
I watched the number climb for a few days, and then it sank in. There was no way any one person could look at all of those. Not for this one role, and not alongside every other role the recruiter is carrying.
I don't know for sure how recruiters get through that kind of volume. My guess is AI helps with some of it. With that many resumes, I'd do the same. But no one knows what I'm hoping for in this role better than I do. That's just the hiring manager's job. AI would know even less.
The AI Move
I didn't make one.
At first it wasn't even a decision. I didn't think to put the resumes into AI. What I thought was simpler: if people took the time to apply, I should take the time to review.
These are people's careers. People get excited about a role. Some of them get hopeful. Yes, some resumes come from bots that apply to everything, but plenty come from people who read the description, felt qualified, and wanted the job. That should be taken seriously by everyone involved in hiring, hiring managers included. The volume is overwhelming, and I wanted to help.
So I spent a couple of evenings in front of the TV, something random playing in the background, laptop on my lap. Click into the application. Get past the title field and the application framework. Open the actual resume. Quick pass.
I'll admit some got a more cursory look than others. The job description asked for 8+ years of program or project management, so scanning job titles cleared a lot of them quickly. VPs were applying. Directors too, which isn't unheard of but didn't make sense for this role. People whose background was front-line work in the food industry, with no program management anywhere.
Then there were titles like Operations Manager. Not an exact match, but close enough that I wanted to read the bullets underneath before deciding. I'm not sure an AI filter would have stopped on that one.
It got faster as I went. After a while I had patterns in my head for which resumes made sense and which didn't. Anything worth a closer look got a note to the recruiter in the recruiting system.
The Shift
Part of me was also curious. I'd never seen what 300 resumes for one role actually look like. It's a mix. Some clearly came from real people who wanted this job. Some made so little sense against the role that I wondered whether a human was involved in sending them at all. I couldn't always tell which was which.
The count kept climbing. A couple of weeks later it was around 600, and that's the first time I thought about AI.
I exported the resumes and gave them to Claude. Before asking for anything, I spent a few minutes telling it what I was actually looking for. Nothing new. Just what I'd already told the recruiter and the other interviewers, in my own words.
It came back with a couple of strong suggestions that I definitely went and looked at. It also came back with flags. A couple of resumes had invisible text hidden inside them, telling any AI system to score them a 10.
I laughed. Then they were out.
I know job hunting is hard right now, and the market is rough. I truly understand that. But you're not going to get a job by cheating the system, and I think most AI tools are wise to that trick by now. It's more likely to backfire.
The Pattern
I used AI all over this hiring process. Interview prep. Taking my rough notes from each interview and working out what I actually thought before manually populating the scorecard. The resume review was the one piece I kept for myself, because it was the piece where I knew the most and any tool would know the least.
Looking back, the order mattered. When I finally handed Claude 600 resumes, I could describe what I wanted in a few minutes. I'd already said it out loud to the recruiter and the interviewers, and I'd spent two evenings noticing what actually made me stop scrolling. That's a lot more than a job description holds.
The Implication
If you're hiring and the pile is big, read enough of it yourself that you can say, in your own words, what makes someone stand out for this role. Then tell your recruiter what you found. It makes their job easier, too.
If you use AI after that, tell it those words first.
What I Will Do Next Time
Next time a role opens, I think I would use AI, just differently. Before anything gets filtered, I'd have it interview me as the hiring manager: what I'm really looking for, which experiences matter, what exceptional looks like, and what is expected in the first 30 days of the role.
Then the filtering runs against that instead of the job description alone. It could catch the Operations Manager I almost scrolled past, the person who wouldn't make it through a plain keyword screen.
It feels like there's room for recruiting software to work this way. Interview the hiring manager, feed that into the system doing the screening, and let the filter be about this specific role.
Comments