Since the start of the LLM (large language model) era, the quality of cover letters used to apply for jobs has increased substantially.
Researchers have found that before LLMs, when candidates needed to write their own cover letters, the average letter on jobs-listing site freelancer.com had a quality score of 5.16 out of 18.
After the introduction of LLMs, when candidates could use AI to write their letters, the average quality score increased to 7.50.
I’d argue this isn’t good news for anyone.
In a world before LLMs, the quality of the cover letter could be used as a proxy for candidate quality. That’s no longer the case – which means recruiters don’t have a way to tell good from bad.
Consequently, the researchers estimate that workers in the top quintile of ability are hired 19% less often and workers in the bottom quintile are hired 14% more often. In other words, recruiters are getting lower-quality candidates and lower quality output.
This situation is making recruiters warier, which is harming workers. The researchers estimate that hiring is 1.5% lower than it would be in a world without AI-written cover letters. Wages are 5% lower than they would otherwise be too.
AI is marketed as a powerful tool for improving productivity, which is undoubtedly true. For example, the same research finds that before LLMs workers applied for an average of 10.40 jobs and the average cover letter was 89 words. After LLMs, those numbers increased to 12.82 and 111 respectively.
But, as the cliché goes, quantity does not always equal quality.
