A world-first research project aims to build software tools that respond to the full spectrum of human traits and demographic factors, rather than just those that dominate. The lack of diversity in software developers and ironing out industry-wide inefficiencies are just two of the issues to address.
How can we empower domestic workers in Bangladesh? How do we increase job applications from females in the Oregon IT industry? How can we help a technology-illiterate grandmother from Melbourne buy wool during a pandemic? And how can we respond to the emotional needs of someone who has recently received a terminal diagnosis?
These are just a few of the questions that software developers and researchers from Monash University’s recently launched HumaniSE Lab have found themselves grappling with in the first two years of their world-first project. The project has the potential to vastly improve opportunities and experiences for a broader range of users, particularly disadvantaged users, as well as increasing the cost-effectiveness of software development across the globe.
“Software was – and overwhelmingly still is – developed by highly educated, relatively affluent, young men,” says HumaniSE lab director Professor John Grundy. “And yet it has become ever more apparent how different we, the end users of that software, all are.”
“It’s not surprising then that a large number of people have problems using it.”
The five-year project aims to put “humans at the heart of software engineering” by creating “intelligent, human-centred future software systems” that take people’s unique qualities – such as age, culture, gender, cognitive ability, emotions and personality – into account.
Professor Grundy has been developing software for more than 40 years and witnessed the problems experienced by users of early software products in the 1980s.
“They’d spent all this money on developing software, only to have to spend a lot more fixing problems that hadn’t been identified early on. It became hugely expensive, as well as leaving many users behind or excluding them altogether.”
The project, funded by the Australian Research Council’s Laureate Fellowship, includes international collaborations with organisations such as Oxfam Australia and Oxfam Bangladesh, the Alfred Hospital, the University of Oregon (US), and Arken University (Germany). The diversity of users these organisations need to reach provides a unique opportunity to test and develop modelling, tools and processes for future software engineers.
Monash University research fellow Dr Jenny McIntosh is leading the digital health area of the research and sees huge potential for reducing health costs by better meeting the needs of a broad range of individual users.
“Digital health is a multi-billion-dollar industry that is increasing exponentially,” says McIntosh. “But it’s also incredibly inefficient. If we can nurture health through simple digital solutions, we could save so much money.”
McIntosh offers the example of a software tool that helps identify risks involved with diabetes – both in terms of preventing diabetes and also managing risks of those diagnosed with the disease.
“If a digital tool can help a pregnant woman, for example, better assess their risk of getting gestational diabetes, they can then take action to minimise that risk. And it’s not just the woman that benefits,” she says. “If people can self-manage their health risks, that’s going to take pressure off the health system as well as empower people to manage their own health.
“A digital tool could also help assess a person’s risk of type 2 diabetes and ideally prevent it, which means a huge amount of taxpayer’s money – that would otherwise be spent on medication, surgery and down the track, perhaps even cardio-vascular disease treatment – can be saved.”
Grundy also points out the vulnerability of many users in the health setting.
“This group is about as diverse as you can get,” Grundy says. “They’re often stressed at the time that they access technology and they might have sharp emotional reactions that poorly designed software can exacerbate.”
Another area HumaniSE sees huge benefits in is the finance sector.
“There is a real worry about the impact the digital divide is having on those less able to access digital finance tools,” says Grundy. “There are vulnerable people who might have older phones that new apps don’t work on. Many developers will make assumptions about users that don’t apply to this part of the community.”
The HumaniSE Lab aims to address this by investing much more heavily in the research stage of software development.
Dr Tanjila Kanij is a research fellow with the project and says that investment in research during software development is a trade-off because although it’s expensive, it pays off.
“If you miss out on a certain type of user, you lose the benefit of reaching that audience. In my opinion, it’s a viewpoint change that outweighs the extra investment.”
Kanij’s research is attempting to reach domestic workers in Bangladesh through a collaboration with Oxfam Bangladesh, one of the international collaborations that is part of the project. These users face challenges with literacy and accessing fair-work environments. Kanij sees the potential for software to empower these workers as significant.
“There are a huge number of domestic workers in Bangladesh but their work is not well recognised,” Kanij says. “As a result, they don’t have set working hours or leave and they’re vulnerable to abuse by employees who have control over their employment and often their living situation. Our goal is to empower these domestic workers with digital technology.”
Two aspects Grundy considers crucial are firstly the effectiveness of tools created by the software, and secondly the diversity of the software engineers creating those tools. The research team includes a high proportion of women, something which is surprisingly rare.
“You might be surprised to know that there was a higher percentage of women working in software development 40 years ago than there is now,” he says.
“There are unfortunately not a lot of women working in the industry,” she says. “Even in university software development courses, women are a minority.”
Australia’s Digital Pulse Report 2021 published by the Australian Computer Society reported that only 25% of workers in the technology sector are women.
According to Dr McIntosh, a culture of diversity is both a way of working and a goal in itself.
“We’re a very diverse team,” she says. “There’s a lot of unconscious bias that we fight against, so we can use our own personal experiences to inform the project.”
Measurement of bias is something that has not been attempted before but which the project aims to create a tool for.
“We’re trying to develop scientific methods to measure and improve diversity through software,” says McIntosh.
One example of these tools is known as GenderMag (gender magnifier), which has already been used at the University of Oregon. The tool helped to create software engineering job advertisements that address uneven gender representation in the workforce, caused by job advertisements being biased towards men.
“One of our recent studies has shown that men have a different approach to seeking and reading IT job advertisements than women,” says Grundy. “If job advertisements are skewed towards this, then the workforce will continue to be dominated by men, which then continues the cycle.”
The lack of diversity goes further than just gender, too.
“We have problems with ageism in the software engineering workforce,” says Grundy. “We have problems with different cultural backgrounds not being appreciated – neurodivergent people and people with physical challenges are not well integrated or supported. This leads to a lack of diversity.”
HumaniSE Lab aims to address all these challenges by diversifying the software engineering workforce and understanding software users more deeply. The project will continue for another three years.
Bron Willis is a freelance environment, science and sustainability writer based in central Victoria.