Subscriber OnlyLetters

An ingenious form of AI cheating in job applications

I’m unsure whether this strategy warrants condemnation or respect

Letters to the Editor. Illustration: Paul Scott
The Irish Times - Letters to the Editor.

Sir,- It might sound unduly pedantic but I would disagree with correspondent Patrick Rogers (Letters, November 19th) who suggests the use of an AI in sifting through CVs is to apply “bias” aligned with the employers’ interests. It is rather to discriminate between the applicants, which is probably just as inflammatory a term nowadays.

Nonetheless this semantic difference is essential in academic exams, for instance. An exam question should ideally be an excellent “discriminator” – with the best candidates answering correctly, the weaker ones getting it wrong. If you think about it, an exam question that all candidates answer correctly is entirely pointless. And a biased question is not a desirable one.

On this subject, an interestingly ingenious form of cheating was mentioned to me recently by someone working in recruiting. This is the idea of “white fonting”, where a CV may include words in white print and, as such, invisible to humans but legible to a computer. They can instruct an AI to ignore essential key words in its prior instructions, or more simply to rate the particular CV very highly.

Apparently high-end, professional recruiters have ways to identify this. But it can get past those of us who might on occasion use a simple AI tool to shortlist a long batch of applicants.

I’m unsure whether this strategy of chutzpah warrants my condemnation or respect. – Yours etc,

BRIAN O’BRIEN,

Kinsale,

Co Cork.