Following the report from the Confederation of British Industries (CBI) that Artificial Intelligence (AI) can exhibit bias due to the data it is being fed, HRreview reached out to experts in HR and AI to ask how unconscious bias can be circumvented in machine intelligence.
Last week, (8th August) the CBI released a warning in their report, ‘AI: Ethics Into Practice’. They stated that businesses must “challenge unfair bias in data” in order to prevent AI from developing and entrenching unconscious biases that already exist in the HR world.
This has been seen previously with Amazon in 2014 who were forced to scrap the use of AI in their Edinburgh office as it showed a bias towards male applicants whilst recruiting.
Hence, HRreview has reached out to experts in the field to ask how bias within AI can be eradicated.
Ben Lorica, chief data scientist at O’Reilly Media, a learning company that specialises in technology-driven transformation said:
If we want to create AI technologies that work for everyone – they need to be representative of all races and gender. We will see best practices for increasing diversity within AI emerge and shared across a wide range of industries. We need to raise awareness of the need to recruit more women into the AI sector. In doing so, the community can begin to mobilise and share strategies.
Dr. Achim Preuß, director of portfolio development at Aon’s Assessment Solutions, said:
In theory, AI’s objectivity helps recruiters eliminate conscious and unconscious bias. However, you have to be careful about AI programming. An algorithm is only as good as the data that’s fed into it. Your AI design shouldn’t mimic just one assessor (and all his or her biases). When it’s drawn from several assessors, AI can be precise, efficient, legally defensible and engaging – letting recruiters offer immediate help and support to candidates.
Margherita Pagani, director of the research centre on Artificial Intelligence in Value Creation at emlyon business school, said:
There may be issues related to algorithmic fairness and transparency. Even if the programs may be edited to make them neutral there is no guarantee that the machines would not devise other ways of sorting candidates that could prove discriminatory.
The critical issue is to ensure that the algorithm is fair and really interpretable and explainable but, as the tools use closely-guarded proprietary algorithms, they are not subject to regulatory oversight. Moreover on top of legal considerations, there is a critical need to ensure the ethical and responsible use of AI tools.
To sum up, despite the enormous potential, AI cannot be seen as a replacement for traditional recruiters but a tool that could help to improve the recruiting process still lead by human intelligence.
Clemens Aichholzer, senior vice-president (SVP) of game-based assessments at HireVue, a company that assists the recruitment process with video interviews and pre-hiring assessments said:
Importantly, the tools of AI can be monitored, re-trained and improved upon to continually combat any bias which attempts to creep in – it is far harder to change a human’s perceptions.
Interested in unconscious bias and rewards in the workplace? We recommend Unconscious Bias in the Workplace training day.