Helena Nimmo, Chief Information Officer at IFS
Society stands at a turning point for Artificial Intelligence. The models we’re building now will shape decision-making, productivity, and social systems for decades. But as we focus on making AI bigger, faster, and more powerful, we can’t overlook a fundamental truth: progress depends just as much on who is shaping it as on what it can do.
Scale without diversity risks amplifying blind spots rather than eliminating them. If we want to build AI that truly works for everyone, we should broaden the perspectives feeding into these systems.
AI isn’t always unbiased, but it can be
AI doesn’t exist in isolation. It’s shaped by the people who build it and the data we feed it learns from. That doesn’t make it inherently flawed. But like any system, its outcomes reflect its inputs. When those inputs lack diversity, the results may fall short.
We’ve all seen what happens when technology isn’t built with everyone in mind: facial recognition that struggles with people of colour, recruitment tools that favour CVs from male applicants, and healthcare algorithms that miss signs of illness in women.
But there’s good news – these aren’t unsolvable issues. Different types of AI offer real opportunities to build more inclusive systems. For example, explainable AI makes biases much easier to detect and correct. These human-in-the-loop approaches make sure that lived experience still plays a role in shaping outcomes.
When we bring in more voices – across genders, ethnicities, cultures, and socioeconomic backgrounds – we bring in more tangible value and experiences. Different perspectives mean different questions are asked. Assumptions get challenged. Edge cases get considered. Systems become more resilient, not because someone created a new algorithm, but because someone in the room had a different lived experience, the whole process becomes more inclusive.
Today, women currently make up less than 22% of AI professionals globally. That’s a structural failure. It means we’re missing a huge range of insight, experience, and judgment from the systems shaping our future.
Thankfully, we have trailblazers like Dr. Timnit Gebru and Dr. Joy Buolamwini (founder of the Algorithmic Justice League) leading the way, calling out bias, demanding transparency, and proving that inclusive AI is powerful. But we need more like them, across all underrepresented groups in this field. We need entire ecosystems of talent from underrepresented communities to shape the direction of travel.
The productivity paradox
With global populations projected to decline from 2080, productivity will become a defining challenge for many economies. This demographic shift means fewer workers supporting larger dependent populations, creating a productivity gap. AI will have to do some heavy lifting to fill the gap.
But that only works if AI if built to serve everyone. If the systems we rely on are designed for a narrow slice of the population, we will limit the gains before they start. An AI workforce that can’t understand or accommodate diverse needs will create technological barriers to participation rather than remove them. We risk developing systems that work well for the dominant group who created them but exclude others from fully participating in the economy when we need everyone’s contributions most.
In a future where human capital becomes increasingly scarce, companies that deploy AI systems capable of supporting diverse users will gain significant advantages. Diversity and inclusion are the foundations for innovation that works at scale.
Too much AI development happens in silos and echo chambers. But inclusive AI? That’s the kind that adapts, evolves, and lifts everyone.
We need to fix this at the source, which means:
- Widening the talent pool by investing in education and pathways for underrepresented communities, including women, ethnic minorities, neurodivergent individuals, and those living outside the major technology hubs
- Changing workplace culture by ensuring diverse voices are actively shaping the direction of travel, rather than just sitting in a cubicle
- Demanding transparency by holding developers and organizations accountable for how their systems are trained, tested, and deployed
- Giving platforms and power to those who’ve lived the realities AI too often misses
None of this is radical. Though it does require some commitment and a shift away from the idea that scale alone is the solution. Diverse teams are better at spotting and fixing biases in AI models than all-male, all-white engineers in Silicon Valley. None of this is about slowing progress – it’s about accelerating in the right direction.
Build it to last
It’s time to move beyond the obsession with “large” AI: large language models, large datasets, and large investments. Intelligence, artificial or otherwise, doesn’t come from size alone. It comes from perspective. From nuance. From context.
We can’t fix what AI doesn’t see. And it can’t see what we don’t teach AI to value.
This is about directing progress. Making sure that AI reflects the messy, diverse, complex reality it is meant to support. Because the alternative is scaling up systems that don’t work for the majority of the population and still calling it innovation.
If we want to build genuinely transformative technology, we have to start with the people shaping it. This can’t be the usual suspects with deep pockets and massive cloud credits, but the full range of human experience. The burden is on enterprises to invite these different opinions into the boardroom.
Let’s stop pretending that “more data” will solve everything. Let’s invite more people to help shape it. The future of AI depends on the perspectives we build into it today.
