Menu Menu
[gtranslate]

How will tech experts prevent AI from adopting extreme bias?

As Artificial Intelligence becomes further interwoven with everyday life, more questions about its input biases emerge.

By 2032, the market value of Artificial intelligence is expected to reach $2,760.3 billion.

This relatively novel technology has become an integral part of our most-used devices, through the incorporation of Siri in our smartphones and computers as well as in our homes via virtual assistants like Alexa.

Now, the use of AI is expanding beyond matters of immediate convenience. It is being applied to many vital and global industries in hopes of improving the efficiency and accuracy of the systems that help them function.

To give just a few examples, AI is now being used in language learning, accessibility, therapy, healthcare, banking, and employment processes.

With essential services making use of AI, concerns about the unintentional biases input into its systems are mounting. Eliminating these biases, tech experts say, requires diversifying the type of people who are doing the programming.

Gender equality in AI programming

The fact is: women are grossly underrepresented in AI tech roles.

According to the most recent data reported by the World Economic Forum, A mere 22 per cent of women make up AI professions globally, with the remaining 78 percent of roles being filled by men. Why is this problematic?

Well, even if male programmers make their best efforts to remain unbiased when working, they are only human. Gender bias can inadvertently be introduced into AI algorithms when the programming team is lacking in diversity.

This bias may result in unfair treatment of individuals based on their gender, reinforcing stereotypes and discrimination – which can have drastic consequences when the technology is used to select candidates for interviews during a hiring process or when making decisions about financial loan approvals.

It is also important to note that having a diverse programming team will enable AI systems to be programmed with a combination of different perspectives, experiences, and backgrounds.

This is essential for creating future AI systems that are inclusive and capable of understanding and addressing the needs of a wide range of users, regardless of their gender.

 

How to keep biases out

Women have spent the last century dismantling systems of oppression.

Having already succeeded in many areas and continuing to chip away at numerous remaining obstacles, women in tech are voicing cautions about how the currently male-dominated AI sector could undo this work.

Meredith Whittaker, who led Google’s AI research in 2006, told Euronews: ‘We live in a patriarchal and misogynistic culture where the positions of power in our world are generally occupied by men, and generally white men in most contexts.’

Whittaker says it is unsurprising, then, that those currently in charge of shaping AI are men.

She continued to say that if our goal is to make AI not just useful but safe for all, the number one priority should be to ensure that those most likely to be harmed by the technology are playing a hand in shaping, teaching, and programming it.

One of the barriers to employing women in AI tech is company culture. The tech world has a reputation for having a ‘boys club’ culture, where women are not typically taken as seriously or rewarded fairly for their accomplishments.

As a result, women are launching their very own tech firms focused on advancing AI – especially for apps that offer assistance with women’s experiences with postpartum depression, menstrual cycles, and more.

Calls for diverse programmers in AI shouldn’t be all that surprising. Marketing teams around the world have learned first-hand about the kinds of mistakes and biases that manifest when creative teams do not include or represent a variety of perspectives.

With AI expected to surpass the capabilities of the human mind in a few year’s time – at least when it comes to efficiency and accuracy – it would be seriously dangerous to have the technology work in favour of a few.

Accessibility