LAHORE: Digital technologies may expand women`s access to information, services and economic opportunities, but access alone does not necessarily translate into control, agency or empowerment, speakers at a session of the AAS-in-Asia conference at the Lahore University of Management Sciences (Lums) said on Saturday.

The second day of the three-day conference featured a session titled `Digital Technologies and the Epistemologies of Gender Research in the Global South` which brought together researchers from anthropology, economics and computer science to examine how digital tools affect women differently across social and economic contexts.

The speakers argued that technologies and research frameworks developed largely in the Global North often failed to account for the social norms, household structures, inequalities and everyday realities that shaped women`s interaction with digital tools in countries such as Pakistan.

Discussing the distinction between access and control, an economics researcher said having access to a mobile phone or financial account did not necessarily mean that women owned or controlled these resources. Women using shared or borrowed phones, for instance, could face restrictionson what services they used because of concerns about privacy, household monitoring or losing control over financial decisions.

The discussion also highlighted how the design of digital tools could overlook context-specific safety concerns. One speaker pointed to navigation applications as an example, arguing that systems designed primarily to identify the shortest route might not take into account the preference of women in Pakistan for routes that were more crowded, visible or well-lit.

Similarly, the panel noted that commonly used digital platforms were often developed without sufficient consideration of how privacy was understood in households where devices and accounts might be shared or monitored.

A major part of the discussion focused on artificial intelligence and women`s health. Dr Maryam Mustafa, who works on technology design and AI for maternal health, highlighted the difficulties of deploying AI systems in contexts where women`s health needs and experiences remained poorly representedin existingdatasets.

She explained that language and cultural context posed significant challenges for AI-based health tools.

`Large language models often perform better in English, while women communicating in Urdu may use everyday expressions rather than formal medical termi-nology. Researchers, therefore, had to work with women, midwives and health professionals to develop a vocabulary that could help digital systems better understand women`s health concerns.

The panel also questioned how the effectiveness of AI tools should be evaluated, arguing that technical or clinical accuracy alone might not be sufficient. Tools intended for specific communities also needed to be assessed by their target users to determine whether information was understandable, relevant and useful to them.

Another key theme was the way research itself is conducted.

Speakers cautioned against relying excessively on rapid, quantitative research models that imposed predefined ideas of concepts such as empowerment and overlook differences within communities.

An anthropological perspective, they said, required longer-term engagement and immersion to understand how people actually used technology and negotiated its effects.

`Such research can reveal differences even among groups that are often treated as a single category, such as rural or low-income women.

The speakers also stressed the importance of interdisciplinary research, noting that effective digital interventions required collaboration among computer scientists, economists, anthropologists, clinicians and practitioners.