The UK can help take the lead in ensuring AI biases are tackled at their source
Recent developments in Artificial Intelligence (AI) are showing just how far we have come with the technology. Take for example the recent showcase from Google, where a telephone call is made by Google Assistant to book a haircut. The technology is now able to understand the nuances of conversation and is quite astonishing at first watch.
With this ability to save time and cut admin, it is no wonder that adoption rates of AI in businesses have grown 60 percent in the past year alone.
However, AI needs to be programmed by a human and in today’s working environment, this can be problematic. Whether conscious or unconscious, biases in the workplace do still exist and this will affect the outcomes of programming the technology.
A recent Lords AI committee report looking into the economic, ethical and social implications of AI recommends that the best way to address these biases is to ensure developers are recruited from diverse gender, ethnic and socio-economic backgrounds. This will help to ensure that the technology is representative of diverse populations.
Workplace Biases Still Exist
Unconscious bias is something that we need to watch out for and arguably, need to actively campaign for change. You only have to look at the recent pay gap reporting to see that gender inequality in the UK still exists.
The reporting found that eight in ten firms pay men more than women. Of those required to submit their employee’s salaries, 3,010 companies have a gap above the national median of 18.4 per cent.
As well as pay inequality, women face unconscious biases on a daily basis. For example, research from Stanford University found that female employees accomplishments are more likely to be viewed as the result of the team, not the individual.
However, businesses are now taking steps to address these issues. For us as a company, our staff are trained to recognise and call out all kinds of biases. Due to this, we have seen a direct positive impact – within a year, management and director level positions held by women rose by 50 per cent.
A focused effort to tackle these inequalities shows that positive changes can be made for women in business. Although, these efforts will take time to permeate. The issue that we will find ourselves with is that AI being programmed today will carry those unconscious biases, placing it behind those positive changes that we are already starting to see. This is something that needs to be regulated.
Avoiding Translating Bias into AI
The truth is, with AI, what you put in is what you get out. It is observing the real-world data that is out there and applying it to its algorithms.
Unfortunately, at the moment, that data is biased. The lack of women in senior positions in the tech industry is a known factor – just 29 percent make up the workforce within scientific research and a study by PwC found that just 5 per cent of leadership positions in the tech sector are held by women. AI that is programmed today will learn that certain careers are overwhelmingly dominated by men.
The AI will take this information and translate it into biases within the algorithms. As such, when a decision is made by AI in the future, it is likely that it won’t be inclusive by virtue of the statistics that it has used, further amplifying the biases already there.
For example, AI is now increasingly being deployed in the recruitment process and may dismiss female candidates for jobs in sectors that it believes is commanded by men.
Gender biases in AI have been observed in many instances, such as where stock photos of women in kitchen and men in fields were not only mirrored by AI, but in fact given a boost. Researchers trained an AI image recognition software to link certain scenes together and found that the AI decided that shopping and washing were things that women do and linked coaching and shooting to men – intensifying the bias that is already there.
To solve this problem and, indeed, stop it from proliferating even further, we have to be able to trust those that are in charge of creating and programming AI. But it is a catch-22, as we know, the technology sector workforce is predominantly made up of young, white males who are unaware of unconscious biases.
This is something that was also looked into by PwC, who found that the gender gap in technology starts at school and continues though every stage of women’s lives.
They found that females studying STEM subjects at school is 64 per cent, compared to that of 83 per cent of males. This takes a steep dive when looking at those at university where just 30 per cent of females opted to study a STEM subject, as opposed to 52 per cent of males. Of the 2,174 students surveyed, a quarter said they have been put off a career as it is too male dominated.
Impact on the Future of Equality
If this notion around STEM subjects continues, it will not only be inherited by generations to come but also by our AI too. Being programmed with any kind of bias will inevitably have a knock-on effect for equality in the workplace.
The UK is in a unique position to help shape the development of AI and to ensure it is free of any form of bias. But tackling these biases at the source is key. If businesses are training staff to recognise unconscious biases, combined with the Governments call to recruit developers from diverse backgrounds, we can ensure that AI will be free from prejudicial decisions. Great strides are being made in terms of equality in the workplace and AI has the chance to help drive this forward.