AI-powered recruitment systems in the UK 'disproportionately' impact ethnic minorities

While concerns about the effects of artificial intelligence systems on human rights, discrimination and ethnic minorities are increasing in the UK, automatic screening systems used in recruitment processes "disproportionately" affect candidates from ethnic minorities.

In the sixth part of Anadolu Agency's (AA) dossier news titled "Discrimination in Europe", concerns about the discriminatory effects of artificial intelligence-supported recruitment systems on ethnic minority candidates in the UK were discussed.

In its report published last month, the Joint Committee on Human Rights (JCHR), consisting of MPs and Lords in the British Parliament, called for new legal regulations, pointing out that existing laws are "insufficient" against the risks posed by artificial intelligence in terms of human rights.

The committee warned that the current legal and regulatory framework is "fragmented and difficult to follow" in the face of the rapid development of artificial intelligence, creating gaps in the protection of human rights.

JCHR pointed out that a "new artificial intelligence law" should be enacted, taking into account the scale and seriousness of the threats that artificial intelligence may cause.

Automated recruitment systems can screen candidates without human sight

Human rights warnings in the UK Parliament are also supported by new research on the use of artificial intelligence and automated systems in recruitment processes.

The research conducted with 2 thousand job seekers by "People Like Us", a non-governmental organization working on ethnic equality in the workplace, and the research company "Censuswide" in the UK, revealed that automatic recruitment software has become increasingly decisive between candidates and jobs.

According to the research, 74 percent of candidates who were rejected for a job in the last two years stated that they suspected that at least one rejection decision was made by the automated system. However, only 15 percent of candidates reported that the employer informed them that it used an automated system to evaluate their applications.

In the research, it was stated that candidates' applications were often rejected without being seen by a human, and some rejection responses were sent before the application period expired or minutes after the application.

Candidates from ethnic minorities are more affected

The study stated that the impact of automatic screening systems was felt especially heavily on candidates from ethnic minorities, and that these candidates applied for approximately one-third more jobs than white candidates, yet they had a higher rate of receiving a rejection response within an hour.

Accordingly, while 41 percent of ethnic minority candidates stated that their applications were rejected within an hour, this rate was 32 percent for white candidates.

In the research, it was stated that 60 percent of job seekers hid or changed some information in their CVs for fear of being screened by automatic systems. This rate increased to 70 percent for ethnic minority candidates.

It was also noteworthy that ethnic minority candidates were more likely to remove information indicating country of birth, religious or cultural identity from their CVs. One in every 20 candidates changed their name.

Candidates try to "beat the machine"

The research showed that job seekers are also starting to use artificial intelligence to bypass automatic screening systems.

While 66 percent of job seekers stated that they tried to "fool" recruitment software, 27 percent said they rewrote their resumes with artificial intelligence tools such as ChatGPT in order to pass automatic screening. 13 percent of candidates stated that they added "keywords hidden in white" to their resumes so that recruitment software would notice them.

According to the research, 31 percent of job seekers stated that they did not apply for a job for which they thought they were qualified because they thought they would be eliminated by artificial intelligence. This rate increased to 39 percent among ethnic minority candidates.

81 percent of candidates said applying for a job now feels like "trying to pass a test for a machine rather than being seen as a human being."

Employers' supervision remains inadequate

The research also raised questions regarding the control of employers using automatic screening systems over these systems.

28 percent of employers using automatic screening said the system does not track whether results differ by ethnicity or other protected characteristics. 82 percent of employers who monitored these differences stated that they detected differences in the results.

Additionally, 30 percent of people involved in hiring decisions shared that they did not know or were unsure how the screening tool they used scored candidates.

According to the research, 75 percent of candidates do not know that they can request a human review against automatic rejection under UK data protection law. While 42 percent of candidates said they wanted to appeal a rejection decision but couldn't find out how, only 28 percent of those who requested a human review said they were able to receive such a review.

Call for transparency and independent audit

"People Like Us" called on employers to clearly communicate the use of automated or artificial intelligence-supported screening systems in their hiring processes. The organization also requested that these systems be independently audited against possible discrimination risks.

People Like Us Co-Founder Sheeraz Gulsher noted that most employers are turning to automation in good faith to cope with increasing application numbers, but it is not a substitute for auditing.

“Automation has a clear role in managing rising application numbers, but there must be checks and balances to ensure recruitment remains fair for everyone,” Gulsher said. made his assessment.

"Artificial intelligence could deepen existing inequalities"

Senior lecturer in Public Policy and Research Methods at the University of Glasgow. Mark Wong's report to the UK Parliament contains similar warnings about the effects of artificial intelligence on human rights, discrimination and racial prejudice.

The report emphasizes that marginalized communities should be centered in the development and use of artificial intelligence systems, otherwise existing inequalities may deepen further.

The report pointed out that the negative effects of artificial intelligence-induced discrimination and prejudice disproportionately affect ethnic minority groups in the UK, and noted that these effects deepen existing social inequalities, marginalization and human rights violations.

The report notes that artificial intelligence systems used in areas such as automatic decision-making, determination of social assistance rights, health services, diagnosis, education, policing, border and immigration practices are prone to racial bias and discrimination.

The report emphasized that technical issues alone would not be sufficient to solve this problem, and warned that if the UK spreads artificial intelligence with its current deregulation-based approach, there is a risk of "automatizing racism" and "further deepening inequalities".

The report included independent auditing, participatory governance and social participation suggestions as solutions. It was also underlined that communities that may be most affected by artificial intelligence should be included in decision-making and control processes.