Why AI-generated recommendation letters sell applicants short

As a principal investigator in a research laboratory that specializes in advanced therapies, I am routinely asked to write recommendation letters for my students, colleagues and associates. I’m also often on the receiving end of such letters from candidates applying for job openings in our group. Over the past year, I have noticed an interesting but concerning development: many of these letters are seemingly being produced not by hand, but by using artificial intelligence (AI) tools such as ChatGPT.
Such chatbots are undeniably remarkable. They can automatically generate letters of reference by drawing on data and patterns to create coherent and grammatically correct compositions. Researchers with limited time, or for whom English is not their first language, can use such aids to make the process of drafting more manageable, ensuring that recommendations are effectively communicated.
Mentoring resources
Nevertheless, as someone who values personal relationships, I find that most of these AI-generated letters lack one key quality, which disadvantages the candidate: the personal touch.
Here, I provide tips for writing an effective recommendation letter and highlight ways to use AI without short-changing the applicant.
A meaningful reference letter requires you to reflect on your personal interactions and experiences with the person concerned. It involves sharing specific examples of instances when you witnessed the candidate’s competencies and behaviours, and incorporating those anecdotes into a narrative that accentuates their strengths and contributions.
AI lacks that foundation, producing text that might be coherent and even accurate, but that lacks emotion or specificity. Instead of a personal endorsement, the result is a letter without passion or subtlety, because it is not based on first-hand experience. And that does the candidate a disservice, because it fails to capture the nuances of their achievements, character, strengths and potential.
Your job in writing a recommendation letter is to go beyond what the candidate has achieved, and predict whether they will thrive in their next role. To do that, you must look beyond cold metrics to consider how they coped with a tough project, developed over time or strengthened the team — insights that can help to contextualize the candidate’s skills and attributes, and perhaps increase their marketability.
For instance, suppose that one of your team members has demonstrated exceptional leadership and project-management skills. You might highlight how they successfully led the team to complete a major project under tight deadlines, detailing the specific challenges they faced. By providing concrete examples, you demonstrate not only their skills, but also their value to potential employers.
When I write about someone’s qualifications, I’m not just sharing facts and figures, I’m also sharing my confidence in them. This authentic endorsement is something that AI just cannot replicate. It’s missing that personal touch and emotional engagement that comes from having lived experiences and memories.
Think about it from the recipient’s point of view: if you couldn’t be bothered to physically write a letter in support of the candidate, who might have been a member of your team for years, why would they want to hire them?
Although ChatGPT and other AI tools possess remarkable capabilities, they are deficient in domains in which human input and discernment are essential. Certainly, you can use AI to polish your text, correct your grammar or turn detailed thoughts into prose that you can then refine. But, in the context of composing reference letters, the deficiency is not a lack of linguistic proficiency; rather, it is the absence of personal connection and authenticity that are derived from tangible human experience.
How do I tell someone that I can’t write them a strong letter of recommendation?
The contemporary world is increasingly digitalized; nevertheless, professional interactions and personal relationships still require human involvement. An effective reference letter is predicated on genuine introspection and a personal recommendation that only you, the writer, can provide. When both requests and replies are driven by artificial intelligence, there is a risk that meaningful conversations could be reduced to mere copying and pasting.
Although letters of recommendation are generally confidential, the candidate might still be able to obtain a copy — for instance, if they are asked to upload the letter themselves to the hiring site.
If, as a candidate, you receive a generic or AI-heavy letter of recommendation, don’t hesitate to reach out to the hiring or funding committee to explain the letter’s limitations and supplement it with materials or personal experiences that highlight your strengths. Alternatively, consider asking for a recommendation from a different referee. Remember, you’ve put a lot of time and effort into your work; it’s OK to ask for a little of your supervisor’s time to support your application.
Whatever your role, don’t let AI ruin an important career opportunity. Both candidates and referees benefit from mastering the art of personalized communication — without relying on algorithmic touches.