When AI makes cover letters free to fake, are employers really screening better, or just screening with less?

This explores what happens to hiring once AI makes the old proof-of-effort signals, like a careful cover letter or a well-written job proposal, nearly free to produce, and what employers are turning to instead. The corpus suggests employers aren't really screening better yet. Mostly they're screening with less information, and some of their workarounds measure the wrong thing.

Start with why writing worked at all. A good application was believable because it cost time and thought to make. Generative AI breaks that link. When the visible output of mental effort is cheap to simulate, it stops certifying the effort behind it, and the same problem shows up in college essays and dating profiles Does cheap AI simulation break the credibility of costly signals?. A Freelancer.com simulation measures the cost. Without written signals, hiring becomes 19% less meritocratic: the strongest workers get hired less often and the weakest get hired more often Does cheap writing weaken hiring based on worker ability?. Game theory has a name for what writing turns into: "cheap talk," which is communication that is free, non-binding and doesn't change anyone's payoff. That makes it easy to exploit, because what someone says no longer has to match what they intend Why can misaligned agents exploit cheap talk channels?.

In practice, employers have mostly answered with filters, and the result looks like an arms race. Greenhouse's survey found 41% of job seekers using hidden prompt injections to get past AI screeners, while 34% of recruiters spend half their week weeding out spam applications. The data supports each side of this loop, but it doesn't show which side started it Are job applicants and employers locked in an escalating AI arms race?. The tools sold as the fix measure speed, not quality. LinkedIn's case for its AI screening rests on recruiter time saved and how many candidates get reviewed, with no data on whether those hires perform better or stay longer Do LinkedIn's AI hiring tools actually produce better hires?. Recruiters are also adopting new cheap signals. Listing AI skills raises interview invitations by 8 to 15 percentage points, but certificates add little over simply claiming those skills. Employers are rewarding a claim they don't check Do AI skills help candidates get more job interviews?.

The more promising answer is to rebuild cost somewhere AI can't easily reach. Kessler splits durable signals into two kinds. One is vouching: a third party puts their reputation behind a candidate, as in a recommendation letter. The other is scarcity: spending limited time on something, like an in-person meeting, shows real interest. Recommendation letters showed measurable hiring benefits. The interest half is still untested Can hiring signals survive when AI makes cover letters worthless?. Both moves shift the signal from what a candidate produces to who stands behind them and what they gave up.

The twist you may not expect comes from inside the workplace. Interviews with 1,250 workers show that once AI is part of the work, people protect cues tied to identity, like their voice and where the work came from. Meanwhile, cues about labor, such as effort, attention and uncertainty, quietly disappear into the finished product Which workplace cues survive AI mediation and which disappear?. So the loss of the effort signal doesn't end at the hiring door. Managers judging finished work face the same blind spot as recruiters reading applications. That suggests the lasting screening signals will be the ones tied to identity and accountability, not to visible effort.