With the surge in use of machine learning and deep learning models for critical decision making in fields like healthcare, social justice and finance, the need to understand, debug and audit these blackbox systems has become pertinent. It has given rise to an entire field called Trustworthy and Responsible AI. Explainable AI is a subfield of this which focusses of explaining the decision making process of these highly complex blackbox systems. Our work concentrates on explanations that are causal and contrastive in nature. Instead of capturing spurious correlations, we aim to capture the underlying causal interactions in the model. We also want our explanations to be contrastive since as humans, we primarily understand outcomes and decision by comparing them to other similar situations. We rarely look at things in isolation, and rely on comparison and contrast to deepen our understanding. We incorporate notions from Game Theory to tackle the feature attribution problem of ascertaining how much difference an input feature makes on the outcome of an instance, and how that input feature value's difference in a given pair of instances affect their outcomes. Our methods address blackbox models that can be either trained on images or tabular data. They can help understand model decision making from the lens of multiple stakeholders- model builders, model auditors or model users. The comprehensive coverage of our proposed methods enable us to produce a toolkit for model explainability that is widely applicable and also generalizable.

Globalization and neoliberal policy shifts have increased market volatility and eroded farmers’ safety nets in India, leading to the world’s largest farmer protests in 2020-2021. Two interrelated trends accompany these shifts: a rapid exit of men from agriculture and intensifying climate change effects. As a result, India, like several other parts of the world, has been experiencing a feminization of farm operation. In the context of an increasingly hostile agricultural landscape, this dissertation explores the implications of an expanded control over agrarian land for female farmers, especially in climate vulnerable districts of India.This research employs a mixed-methods approach. For the qualitative segment of my research, I conducted 45 focus groups with 583 women farmers as well as 18 semi-structured interviews with journalists, NGO representatives, government officials, and farmers in five climate vulnerable districts (Chitrakoot, Parbhani, Beed, Dharashiv, and Latur). These districts have three types of male exit from agriculture: outmigration, farmer suicides, and a greater focus on nonagricultural activities in the village. I supplement my qualitative fieldwork insights with my own survey administered to 300 farm households in Chitrakoot and an econometric analysis of 9698 rural cultivator households using the nationally representative India Human Development Survey 2 (IHDS 2) for external validity. This dissertation has three key findings. First, female headed farmer households tend to be negatively selected: they are less wealthy and own smaller areas of land than male headed farmer households. Second, land (both quantity and quality) is an extremely important determinant of women’s farming experiences and significantly mediates women’s access to and use of other complementary agricultural inputs such as water, labor, and capital. Third, patriarchal norms exacerbate negative selection issues and further reduce female farmers’ access to vital agricultural inputs such as labor and water. Therefore, while a vast literature outlines the benefits of land rights for female empowerment, I argue that the circumstances under which land rights are received matter (type of male exit, land characteristics, patriarchal norms, and climate change). Feminization of farm operation grants women farmers operating or ownership rights to agrarian land but because of the above factors, the women I study are not necessarily better-off as a result of these land rights. My research also advocates for a more gender sensitive agricultural policy approach that takes into account restrictive socio-cultural norms that women farmers face as well as the need to design policies tailored to the socio-economic background of the woman farmer instead of treating women farmers as a homogeneous group.