How AI Might Hinder a Diverse and Inclusive Workplace

Artificial intelligence (AI) is being used in countless ways in the modern world, from self-drive cars to virtual assistants and open-source software. So it’s no surprise to see AI increasingly being employed to improve matters in the world of business. One such area is that of recruitment. AI can dramatically reduce the amount of time the talent acquisition process can take, but as we’ll see, it can prove harmful for those whose business goals include improving diversity and inclusion in the workplace.

What is a diverse and inclusive workplace?

A diverse and inclusive workplace is one that embraces all human points of difference. These may include gender, race, age or lifestyle preferences. A truly diverse workplace will have people from all genders and races at all levels of the business, from, say, frontline staff to the boardroom.

Given notions of fairness, combined with the fact that diverse and inclusive workplaces bring better results for a business, ensuring your recruitment processes do not discriminate against any group in society is vital.

Gender equality solutions: the use of AI in recruitment

As any business leader knows, the talent acquisition process can be a lengthy one. Companies may receive hundreds or even thousands of applications, meaning a large amount of time is consumed by the shortlisting process alone. AI can be utilised to quicken this process up, by analysing huge amounts of data in a much shorter timescale than it’s possible for a human to achieve. It also allows TA consultants to focus their attention on more meaningful, qualitative work.

AI works through this program by using pre-programmed algorithms. These may be keyphrase-led, with the digital process shortlisting candidates through a series of specific words or phrases. It may also be used to check a candidate’s online presence and build up a personality profile.

Furthermore, many support the use of AI in recruitment for reasons other than as a time-saving tool. Given that we as humans can carry unconscious bias without even realising it, a machine should be able to sift through candidates minus any inbuilt human prejudice. It’s a great theory, but as we’ll see below, how it works in practice may be very different indeed.

The potential for AI bias in talent acquisition

There’s an old saying that computers and robots are only as clever as the person who developed them. Even with concepts such as AI and machine learning, which involve computers using data to grow more knowledgeable over time, much depends on the person who initially programmed the algorithms to be used.

All of us carry elements of bias in our brains, whether conscious or otherwise. Societal stereotyping may lead a programmer to unwittingly create an AI module that adheres to the same stereotypes.

Such a process has already been found within the AI of some seriously-huge organisations. In 2018, Amazon discovered that the AI-based recruitment tool it was using was biased against women. No matter how intelligent an AI model may appear, it can only act on the data it’s been fed.

Remember, AI is utilised to make human lives easier. Within the sphere of recruitment, using machine learning was judged as a method to make the system fairer. Regarding the creation of more diverse teams, removing unconscious bias from the recruitment process is a must if we are to overturn centuries of anti-female discrimination. But cases such as the Amazon declaration demonstrate how easy it is for bias to be transferred from human to machine, achieving the very opposite results to those intended.

Using AI to remove unconscious bias

Let’s not forget – one of the main reasons why AI has been introduced in the recruitment process was to remove any human bias from the process and so be one of many ways to improve gender equality in the workplace. So how can we ensure that cases such as the Amazon incident remain a minority.

1.Refresh your knowledge

Given that unconscious bias happens outside of our control, business leaders need to make a concentrated effort to uncover their own inbuilt prejudices but don’t try to do this alone. Set up a diversity think-tank, and make sure it’s diverse in its make-up, then garner opinions on your job advertisements and interview processes.

Those from different backgrounds may be able to identify areas of unconscious bias in your language, which appear neutral to you. Find our article on increasing gender diversity in business with gender equity for more information.

2.Start with a diverse development team

The best solutions are often the simplest. If you’re looking to create a diverse and inclusive workplace and want to utilise the benefits of AI, then make sure your AI-development team is diverse. Leaving it to one person will likely lead to instances of unconscious bias being found within the algorithms. Whereas people from different backgrounds bring different ideas and perspectives, meaning there’s a greater chance of any bias being spotted.

3.Don’t rely on AI

We all have an opinion on the rise of machines. Some business leaders are wary, others all-embracing. Whatever your own take on the technology, it’s important to remember that computers are just another tool that can potentially help us, and not a complete replacement for the humble homo sapien.

Robots can be programmed to do incredible things, but without us humans, they’re nothing. Just like raising children, they need careful assistance if the results are to meet our expectations.

For further reading, see our post on how to remove gender bias from your recruitment process.

                                                  ________________________________________

AI has the potential to revolutionise our workplaces, rolling back centuries of gender-related oppression by removing unconscious bias from the equation. But a lack of diversity within your development team can easily hinder your attempts to create a diverse and inclusive team.

A careful balance is needed between employing robots to make our lives easier, and not losing sight of where the robots’ thought processes have come from. Should robots fail to deliver the results we’re looking for, we’ve only got ourselves to blame.