From Tamiliniyaa Rangarajan

Data is fundamental to scientific innovation…

From identifying research gaps to piloting new technologies and collecting user feedback, data not only fuels science anticipation but also tangible breakthroughs. Nowhere is this more evident than in the field of artificial intelligence, where datasets are crucial to powering machine learning algorithms. Be it powering facial recognition technologies (FRT) at airports scanning passengers or equipping large language models (LLMs) to predict the next word in a sentence, large and diverse datasets allow algorithms to identify patterns, make predictions and improve the accuracy of their outputs. As advances are made in the field of quantum computing – producing compatible hardware, correcting errors and mitigating noise – we inch closer to another revolution in data analytics. 

Quantum systems, leveraging principles such as superposition and entanglement, can process vast quantities of information in parallel, breaking through the limitations of classical computing. Experts at GESDA foresee its ability to solve complex calculations, such as using quantum simulation to investigate the nature of physical materials and decrypt existing cryptographic protocols. Yet before this anticipatory quantum future becomes a reality, there is the need to address one critical challenge—the quality and the representativeness of the data we rely on presently. 

Data, like any other human-generated resource, is shaped by the society that produces it. As a result, contemporary datasets, either consciously or unconsciously, often encode deep-seated historical and social inequities. Take for example, Joy Buolamwini and Timit Gebru’s analysis of commercially available facial analysis datasets.[i] Their research demonstrates that because these datasets are overwhelmingly composed of lighter-skinned individuals, algorithmic accuracy of FRTs when classifying darker-skinned women is significantly lower. This reflects real-world implications for fairness, safety and accountability as well limits the reliability of emerging technologies and their diverse applications. This risks deepening existing digital divides. 

When breakthroughs in quantum computing are achieved over the next five to ten years, their increased computational capacity will only exponentially increase the rate at which systemic biases are encoded. Drawing from Professor Erika Kraemer-Mbula’s comments about innovation economies at the 2025 GESDA Summit provides an interesting extrapolation. Much like how the concentration of wealth will lead to reduced overall demand – shrinking the market for innovation – the concentration of data representation to very specific demographics will limit the applicability of emerging technologies. When only a small percentage of the global population is represented in the datasets that train AI, and eventually feed into quantum-enhanced systems, these technologies will lose their legitimacy. 

As a result, advanced computing power alone will not translate into equitable technological benefits. If intentional interventions to improve data quality by diversifying sources and ensuring representation are not taken, scientific innovation will remain uneven—furthering existing inequalities instead of resolving it in the first place. Moreover, representative data is not just a technical necessity, but is also crucial to ethically capturing the human experience. With the increasing integration of emerging technologies into everyday life, poorly trained systems can deny individuals access and mobility in dangerous and demeaning ways.

Science anticipation must then closely consider developments in data analytics to ensure that their predictions are impactful.

[i] Joy Buolamwini and Timnit Gebru, “Gender Shades: Intersectional Accuracy Disparities in Commercial Gender Classification,” PMLR, January 21, 2018, https://proceedings.mlr.press/v81/buolamwini18a.html