Talk Title: Intelligence is Social

Many of the research in AI is devoted to create systems that perform tasks or functions for which we think humans need intelligence. Famous are things like playing chess or go, but classification and decision making are also among the popular ones. However, most of these endeavours are geared towards solving a problem and/or optimizing a solution individually. In real life intelligent behaviour is determined by our relationships and interactions with other people.
The consequences of taking this serious for AI research are twofold. If we...

Talk Title: Designing Technology for BIPOC Joy and Flourishing

Research and design in human-computer interaction centers problem-solving, causing an unintended downstream effect of framing work centering marginalized communities predominantly from the lens of deficit and damage. In this talk I will present "designing for BIPOC flourishing" - a design framework that centers values such as joy, play, pleasure, rest, and cultural heritage for Black, Indigenous, and communities of color as well as a...

Talk Title: Doing Some Good with Machine Learning

This is the story of my assorted attempts to do some good with machine learning. Through its telling, I’ll highlight several models of organizing social good efforts, describe half a dozen social good problems that would benefit from our community's attention, and present both resources and challenges for those looking to do some good with ML.

Speaker: Lester Mackey, PhD

...

Talk Title: The moral salience of the algorithmic lens

Our world today is filtered through the algorithmic lens -- a vast assemblage of algorithmic tools and data that mediate our experience of life, work, and health. This lens reveals and distorts: it forces the world into forms that are machine-readable, and in doing so renders some aspects sharper, and makes others invisible.

The algorithmic lens requires precision, and data. It seeks to optimize, and does so at scale. But to render ourselves amenable to the algorithmic lens, we make choices --...

Talk title: Reimagining Tech Design: Crafting Inclusive Futures Through Alternative Narratives

Talk Abstract: Technology permeates nearly every facet of our lives—it shapes our health and well-being, transforms our work and workplaces, and revolutionizes transportation and education. However, this digital transformation has been far from equitable. Studies reveal that job seekers and entrepreneurs from lower-income and underrepresented ethnic groups are often excluded from the...

Talk Title: Generalization vs Specialization in Computer Vision for Ecology

Talk Abstract: There are explicit tradeoffs between generality and specificity in computer vision methods with respect to designing impactful end-to-end solutions for ecological challenges — computer vision systems that are general purpose but optimal for each stakeholder, taking into account cost, human effort, and risk. We explore these tradeoffs across several...

Talk Title: Participatory AI/ML for Grassroots Data Science

Critiques of big data, artificial intelligence and the political economies of platforms have been mounting in recent years, with scholars exposing their biased algorithms, extractivist logics and discriminatory outcomes. How, then, can those of us who work with AI, data science and technology use such tools in the service of equity, justice and liberation? Drawing from my forthcoming book Counting Feminicide: Data Feminism in Action (MIT Press, 2024), I describe a case study of building participatory...

Talk title: A unified framework for robustness, fairness, and collaboration in machine learning

“Pervasive needs for robustness, multi-agent collaboration, and fairness have motivated the design of new methods in research and development. However, these methods remain largely stylized, lacking a foundational perspective and provable performance. In this talk, I will introduce and highlight the importance of multi-objective learning as a unifying paradigm for addressing these needs. This...

Talk title: Computational Challenges in Environmental Epidemiology: Embedding Spatial Data Science, Mobile Health Technology, and Deep Learning into Prospective Cohort Studies

Abstract:  The places in which we live, work, play, and age influence our health behaviors, our mental health, our cognitive function, and our chronic disease risk. However, the majority of research on spatial factors and health has relied on residential addresses to assign exposure, questionnaire data to measure health behaviors and health outcomes, and coarse and nonspecific indices to estimate...

Talk Title: A pragmatists' guide to machine learning for Earth observation imagery

Abstract: Our planet is constantly changing as a result of ecological and socioeconomic interactions across a variety of spatial and temporal scales. A growing number of Earth observation datasets provide regularly repeated synoptic measurements that enable us to map and understand landscapes as the dynamic systems...

Talk Title: Using AI to Advance Mobile Digital Health in Resource-Constrained Environments

Abstract: The emergence of ubiquitous mobile computing devices and Artificial Intelligence (AI) algorithms is transforming patient care, hereby making it more scalable and accessible. This is of particular importance in resource-constrained environments (RCEs) where access to healthcare technologies or expertise may be limited. The sensing modalities (e.g., camera/gyroscope) of mobile devices are capable of capturing and storing data that can then be...