AI: whatever your opinion on it, while it can be easy to get caught up in the commotion
related to how it will impact jobs, many have overlooked the impact of underlying bias
existing within the technology.
While much has been written about AI bias stemming from training data, the real threat is a
more insidious problem: invisible bias rooted in socio-economic inequality. No single
organisation holds the answer, but Scotland’s unique ecosystem is perfectly positioned to turn
this challenge into a global opportunity.
The high stakes of legacy bias
As with any new technology, it can take time to work out the bugs that may exist in it. This is
no different when it comes to AI.
But when it comes to AI bias, we tend to focus on what’s easiest to measure things like
gender, race, or age. Of course, each of these markers are hugely important, but bias also
hides in much more subtle places, such as hidden or silent disabilities and socio-economic
inequality caused by the education lottery that many experience.
I grew up in a community where schools were under-resourced and results were poor – and
remain poor. During that time, digital literacy and AI were nowhere near on the agenda. If AI
systems are trained on data that assumes that everyone has the same digital access, stability,
and educational opportunity, then whole groups of people become completely invisible,
excluded from how the systems makes predictions or decisions.
That’s a real problem for Scotland because so much of our economy and public sector relies
on data drawn from legacy health, social care, education, and finance systems that all reflect
decades of structural inequality. If we’re not careful, AI won’t just mirror those inequalities, it
will amplify them at scale and speed.
Co-design: The Scottish advantage
Despite what might be going against it, there is significant opportunity in Scotland to be the
country where biases become a thing of the past and ethical AI thrives.
Sure, tackling this won’t be something that can or will be changed overnight, but it is
important that as a collective we shift our mindsets from looking to correct bias, to co-
designing systems that address it. After all, ethical AI is not an add-on; it must be built into a
foundation and that all starts with who is in the room; the voices who will shape the datasets
and whose experiences are reflected in the outcomes.
For Scotland, we’re small enough and connected enough to do things differently when it
comes to embedding ethics and inclusion into our AI ecosystem from the ground up, and to make sure that the technology we build reflects all of Scotland, not just the parts of it that are
easiest to capture in data.
This starts by looking at the interconnected relationship between industry, government,
academia, and wider society. Each group brings a crucial perspective: Industry drives
innovation and commercial outcomes; the Public Sector provides necessary policy
frameworks and accountability; Academia contributes research and critical thinking; and
Civil Society offers lived experience and an understanding of real-world impact. Only when
these diverse voices collaborate early and often will we be able to design AI systems that are
genuinely fair, transparent, and trustworthy.
From checkbox to culture
But the challenge isn’t just technical, it’s cultural and structural. Many organisations still
don’t have the means or the mindset to look for bias in the first place especially when it’s
buried deep in the assumptions behind the data. It takes more than a fairness tool or a
governance framework to fix that. It takes awareness, data literacy, and leadership that
actively invites diverse perspectives into the design process.
What is important to consider is that AI accountability starts long before any deployment. It
starts with ownership and transparency over who has originally designed the technology and
what assumptions have been embedded within it. And if that AI has been co-designed by a
cross section of parties, then we are onto the start of something positive.
Even still, when it is time to implement AI, it is important for organisations to question these
assumptions. Not only will this practice ensure that users will be more invested in the use of
AI, but they will also be able to trust the results that it delivers. The goal is to make
accountability a shared cultural value, and not just a governance framework tick box exercise.
This means leaders must activity empower teams to ask: “Should we?” before asking “Can
we?”
At the end of the day, ethical AI isn’t a competitive advantage; it’s collective infrastructure.
When Scotland’s universities, startups, government, and communities align, we don’t just
build better AI – we build public trust, and that’s what really sustains innovation.
Thi is a guest post from Scotland Women in Technology (SWIT) board member, Margaret Totten, who advocates for the co-design of AI systems from the outset and how Scotland could become a leader in this area. It brings together diverse voices from across industry, the public sector, academia and society.
Find out more about them at their website scotwomenintech.com
