I prost NLP algoritam na osnovu CV-ja zna pol kandidata, navodi istraživanje

https://www.unite.ai/job-applicant-resumes-are-eff... link-img
avatar

Algoritam je bio u stanju da prepozna pol kandidata sa 70% uspeha čak i u slučajevima gde je dosta sadržaja uklonjeno. Deluje da u ovom trenutku ni postojeći modeli veštačke inteligencije, ni moderni HR trendovi koji žele da smanje diskriminaciju nisu u stanju da eliminišu predrasude koje su utkane u jezik, i čak mogu da ih učvrste, jer se korišćenje algoritamskih metoda ističe kao nepristrasno ili neutralno rešenje. Možda nam najpre treba nov način zapošljavanja, koji ide mimo CVja.

Iz članka:

Researchers from New York University have found that even very simple Natural Language Processing (NLP) models are quite capable of determining the gender of a job applicant from a ‘gender-stripped’ résumé – even in cases where machine learning methods have been used to remove all gender indicators from the document.

Following a study that involved the processing of 348,000 well-matched male/female résumés, the researchers conclude:

‘[There] is a significant amount of gendered information in resumes. Even after significant attempts to obfuscate gender from resumes, a simple Tf-Idf model can learn to discriminate between [genders]. This empirically validates the concerns about models learning to discriminate gender and propagate bias in the training data downstream.’

The finding has significance not because it’s realistically possible to hide gender during the screening and interview process (which it clearly is not), but rather because just getting to that stage may involve an AI-based critique of the résumé with no humans-in-the-loop – and HR AI has obtained a besmirched reputation for gender bias in recent years.

1

ULOGUJ SE I DODAJ KOMENTAR


Registruj ili prijavi.

VIDI POPULARNE POSTOVE


2
4
avatar Anon pre 5 dana