Too old for a job interview? 'Age discrimination is everywhere, but we’re often unaware of it'
Ageism
How can we hack prejudice about age and ageing out of the recruitment process? That was the question data scientists, psychologists and HR professionals tackled during the AnyAge.AI hackathon. 'AI can actually be used to increase fairness in recruitment.
Anyone applying for a job later in life risks falling prey to ageism: negative prejudices and stereotypes about ageing. In the workplace, this might take the form of assumptions that older employees are slower to pick up new technological skills, or resistant to changes in working practices.
Fresh energy
Recruitment processes often rely on software powered by artificial intelligence to pre-select CVs and cover letters, which means that older candidates are frequently, and often unintentionally, filtered out. This is partly down to the job adverts themselves, says Ittay Mannheim, a researcher in Methodology and Statistics who specialises in ageism and digital technology. 'Some wording is clearly geared towards younger applicants- words like "fresh energy" and "dynamic", for example.' HR managers are often unaware of the implicit bias in such language.
Fair recruitment process
To figure out how recruitment could be made fairer, Mannheim and his colleague Rüya Koçer organised the AnyAge.ai Hackathon in late June. Nine multidisciplinary teams, made up of experts in psychology, data science and HR, set out to find solutions to ageism in selection procedures. 'What made this hackathon unique was its participatory approach: we conducted interviews with HR managers beforehand to formulate our challenge, and they were also part of the jury. We also consulted with a steering committee several times in designing the challenge.'
'Remove age from CV's, this helps prevent unconscious bias during selection'
'As Recruitment Adviser, Labour Market Communications Specialist, and Policy Adviser for Labour Mobility within the HR Policy team, we are engaged every day with recruitment and selection, as well as with themes such as diversity and inclusion. When this AI Hackathon came our way, it immediately appealed to us. We wanted to learn more about it, share our ideas, and contribute to possible solutions. In a multidisciplinary team, we were able to draw on everyone’s expertise. What we will take with us to the next selection round is this: remove sensitive information, such as age, from CVs. This helps to prevent unconscious bias during selection. Thanks to increasingly sophisticated algorithms and tools, it is becoming possible to design a selection process that does not discriminate, but instead values talent for its true worth.'
Michèlle van Wijk, Daniëlle van Rijk and Brenda Prins were members of the winning team of the hackathon
Effective solutions
'The participants were given a dataset to train their algorithms on, and worked with two vacancies — one junior, one senior — and ten applicant CVs. They then applied their own prototypes to ultimately produce a list of the ten most suitable candidates.' The challenge for the teams was to devise the most effective strategies for reducing AI bias.
In the end, three teams won prizes of €2,000, €1,000 and €500 respectively. 'The jury looked at whether the solutions made the recruitment process fairer, whether the team had already built a workable prototype, and how original the ideas were.' One team, for example, proposed a system to revise so-called ‘borderline cases’ — CVs that weren’t clearly suitable or unsuitable. Another suggested using AI to advise HR professionals on inclusive language in job adverts. The teams also talked about the necessity to look beyond algorithms, and to ask how awareness of bias could be raised more broadly within.
Awareness
'Ultimately, that’s exactly why we organised the hackathon,' says Mannheim. 'To bring more awareness to this issue. In that sense, the hackathon itself could also be seen as an intervention aimed at change.' And that change is sorely needed, the researcher argues. 'We’re less aware of age discrimination than of many other forms of bias. Yet it pervades our entire society: healthcare, technology, media. If we don’t address it, eventually everyone will suffer the consequences.'