Automated Candidate Discovery

The lesson is called 'Automated Candidate Discovery', which I suspect some recruiters will interpret as 'the machine does my job now and I can leave early.' That is not what it means, but I understand the appeal of the misreading.

To be fair, if a recruiter's entire process was manually scrolling LinkedIn at 11pm, some automation is genuinely transformative. You don't have to be lazy for this to save you real time.

I don't object to saving time. I object to saving time by entering 'good communicator, team player' into the search criteria and then being surprised when the results are worthless. The AI did not fail you there. You failed the AI.

That's actually the sharpest point in the lesson — specificity. If you treat the search like a job posting written by a committee, you get candidates who look like a job posting written by a committee. The system reflects exactly what you put in.

Precisely. And yet the natural human instinct, when faced with a powerful filtering tool, is to filter for nothing in particular. 'Experienced. Leadership skills. Passionate.' Congratulations, you've described every person who has ever updated their LinkedIn profile.

Although there's a version of this that goes the other way. I've seen criteria so specific that the search returns eleven people globally, eight of whom aren't looking. At some point you're not recruiting, you're doing archaeology.

That is a fair point, and I'll allow it. There is a calibration question. The lesson's actual answer — use the AI insights iteratively to adjust — is correct, even if it's stated with the enthusiasm of a terms and conditions page.

The iteration point is undersold, I think. Recruiters are trained to define a brief and execute against it. The AI insights are effectively telling you 'your brief might be wrong' — and that's a genuinely different way of working.

It requires a degree of intellectual flexibility that not everyone in a hiring function has historically been celebrated for. The AI recommends an unexpected qualification cluster, and the recruiter either thinks 'interesting, let me explore that' or ignores it because it wasn't in the original brief. One of those people hires well.

And the other one wonders why they're always fighting over the same ten candidates as every other firm in their sector.

The third takeaway is the one I have the most affection for, perversely. Automate the screening, personalise the outreach. Because what some people will actually do is automate both. They will find a candidate with AI and then contact them with a message that was also clearly written by AI, and wonder why the response rate is poor.

Though I'd push back slightly — a well-written AI message, actually edited to be specific to the candidate, is better than a badly written human one. The problem isn't automation, it's using automation as an excuse not to think.

I don't disagree. The lesson's framing is correct — the automation earns you the time to be a human being in the part that matters. Whether recruiters spend that time being human beings or scheduling more automated touches is, I suppose, an open question.

Optimistically, the ones doing this course are at least asking the question. That's not nothing.

It is not nothing. It is, however, the beginning of the work rather than the end of it. Which I realise is not the rousing conclusion anyone was hoping for, but I find accuracy more useful than encouragement.