I want to note many of the things you outline as problems above, such as poor hiring decisions and poor matching--these were problems long before LLMs broke into the scene. Companies outsourcing their hiring processes to recruiters broke the system, because the incentive of recruiters is misaligned with those companies: recruiters maximize revenue by shoveling shit, not by tailoring perfect matches.
All AI did, at least on that front, is accelerate the process.
You're talking about the bottom end of the ecosystem. That was opt-in, per-company. Well-run and average orgs could still rely on the candidate's words.
The "market for lemons" comparison is quite apt, as it seems to risk hurting the finding of the better candidates.
I find this fascinating and great learning about that paradox. I struggle these days with the notion about saving time but so few are saying what those savings do. If I saved time doing x, what could I do with that time? The completion of the task faster is great but what more worthwhile endeavor did it unlock? If we got rid of all the data chumming toil, do we take that time to deeply think creatively about something else? Do we get something else done that would have gone unrealized previously?
The common pattern I see is an AI-amplified arms race: a negative-sum game where individual actors are acting optimally, but the players in aggregate are worse off. A prisoners' dilemma.
Definitely. The job candidate ~should~ use AI to draft a cover letter and the company -- by virtue of the volume they're receiving -- likely needs automation to sift through it. But the system doesn't serve anyone well, except bottom-quartile candidates who now have a greater chance to get hired.
I was thinking it doesn't make sense to have to pay for successful cold calls. It'd be better to charge a small amount for every failed CC to discourage bad ones, pool that money and then distribute it in larger slugs to the charities selected (or individuals?) each time someone responds to a CC.
Yes.
I want to note many of the things you outline as problems above, such as poor hiring decisions and poor matching--these were problems long before LLMs broke into the scene. Companies outsourcing their hiring processes to recruiters broke the system, because the incentive of recruiters is misaligned with those companies: recruiters maximize revenue by shoveling shit, not by tailoring perfect matches.
All AI did, at least on that front, is accelerate the process.
You're talking about the bottom end of the ecosystem. That was opt-in, per-company. Well-run and average orgs could still rely on the candidate's words.
The "market for lemons" comparison is quite apt, as it seems to risk hurting the finding of the better candidates.
Lovely bit of writing and thinking - thanks for sharing!
I find this fascinating and great learning about that paradox. I struggle these days with the notion about saving time but so few are saying what those savings do. If I saved time doing x, what could I do with that time? The completion of the task faster is great but what more worthwhile endeavor did it unlock? If we got rid of all the data chumming toil, do we take that time to deeply think creatively about something else? Do we get something else done that would have gone unrealized previously?
Great article!
The common pattern I see is an AI-amplified arms race: a negative-sum game where individual actors are acting optimally, but the players in aggregate are worse off. A prisoners' dilemma.
Definitely. The job candidate ~should~ use AI to draft a cover letter and the company -- by virtue of the volume they're receiving -- likely needs automation to sift through it. But the system doesn't serve anyone well, except bottom-quartile candidates who now have a greater chance to get hired.
AI generated outputs are alot like junk food.
No mine
Really interesting.
I was thinking it doesn't make sense to have to pay for successful cold calls. It'd be better to charge a small amount for every failed CC to discourage bad ones, pool that money and then distribute it in larger slugs to the charities selected (or individuals?) each time someone responds to a CC.