A coalition of leading research bodies has launched Doing AI Differently, an initiative that challenges the way artificial intelligence is developed and understood. Led by The Alan Turing Institute in partnership with the University of Edinburgh, the Arts and Humanities Research Council (AHRC-UKRI), and the Lloyd’s Register Foundation, the project calls for a shift toward a more human-centred, interpretive approach to AI.
For decades, AI has been treated primarily as a mathematical tool—an engine for crunching numbers and producing results. But according to the team behind the new programme, this perspective misses the bigger picture. AI outputs, they argue, are more like cultural artifacts than spreadsheets—comparable to novels or paintings created without true comprehension of their meaning.
This lack of “interpretive depth” becomes a critical flaw when nuance and context are essential, says Professor Drew Hemment, Theme Lead for Interpretive Technologies for Sustainability at The Alan Turing Institute.
“The system doesn’t really understand what it’s saying,” he explains.
One key problem is the so-called “homogenisation” of AI. Most AI systems today are built using similar architectures, which means they share the same blind spots, biases, and limitations. Professor Hemment compares it to “every baker in the world using the same recipe”—you might get consistent results, but you also get a lack of diversity, innovation, and adaptability.
The consequences of such uniformity are not theoretical. Social media, for example, began with simple goals but evolved into a global force with unforeseen societal impacts. The Doing AI Differently team hopes to prevent AI from following the same trajectory.
Their proposed alternative is Interpretive AI—systems designed from the ground up to embrace ambiguity, multiple perspectives, and cultural context. This approach would allow AI to offer more than one “right” answer, mirroring the complexity of human thinking. It would also encourage exploration of new AI architectures that move beyond today’s standard designs.
Rather than replacing human decision-making, the vision is for AI to work alongside people in “human-AI ensembles,” combining human creativity and judgement with AI’s computational power to tackle challenges such as healthcare, climate action, and public safety.
Ryan Daws
The potential applications are tangible. In healthcare, an interpretive AI could help capture the full patient story—beyond symptoms and test results—leading to more personalised care. In climate policy, it could bridge the gap between global scientific data and the cultural and political realities of local communities, fostering solutions that are both effective and culturally relevant.
The project also marks the start of an international funding programme, bringing together UK and Canadian researchers to advance this vision. But time is of the essence.
“We have a narrowing window to build in interpretive capabilities from the ground up,” warns Professor Hemment.
For partners like the Lloyd’s Register Foundation, safety remains the guiding principle.
“Our priority is to ensure future AI systems, whatever shape they take, are deployed in a safe and reliable manner,” says Jan Przydatek, the foundation’s Director of Technologies.
At its core, Doing AI Differently is about more than advancing technology—it’s about ensuring AI evolves in a way that reflects human values, respects diversity, and supports society’s most pressing needs.
