Overview

Technological advances, such as AI, are seismic forces shaping the 21st century. However, vulnerable groups are getting left behind. The speed, scale, and top-down nature of technology require a deliberate counter-approach empowering vulnerable groups to not only be better users and protect themselves from being victims of technology, but also be creators of technologies.

Open, online competitions are an approach grounded in behavioral research that can build technical capacity in cost-effective ways in all types of communities.

CiML (Challenges in Machine Learning) is a forum that has been bringing together workshop organizers, platform providers, and participants since 2014, to discuss best practices in organizing competitions and new methods and application opportunities to design high impact competitions. Following the success of previous years' workshops, we propose to reconvene and discuss creative opportunities for broadening our community and using competitions for social good.

With 1.3 billion people classified as multidimensionally poor (half of whom are children), ML competitions can be a powerful approach to not just increasing AI literacy but also for: 1) building self-efficacy; 2) upskilling to match 21st-century needs, and 3) expanding capabilities for the most vulnerable groups.

The COVID-19 pandemic has brought to light additional barriers for society’s most vulnerable communities. Technology access, which was previously a luxury for many underserved communities, now is a necessity for vulnerable families to continue their education and widen their life choices (UN Development Programme, 2019). This problem, however, runs deeper than simply providing access. For underserved communities, most remote education offerings are not sufficient to overcome the lack of knowledge, social capital, and deeply entrenched systemic inequities. The pandemic’s impact on education is especially hard for girls and mothers, who are expected to take on additional responsibilities of family caregiving, hygiene, and health.

Within this context, technology/AI capacity building must improve if nations are to meet economic health, education, and general well-being needs during and beyond COVID-19. We need to ensure that communities are able to access the educational resources they need to build new skills, but also become drivers of solutions to pressing problems they are facing and have intimate understanding of. If we do not all come together to support vulnerable communities, then the progress humanity has made over the past 50 years in education, equality, and health may get wiped out.

For this seventh edition of the CiML workshop at NeurIPS our objective is threefold: (1) We aim to foster diversity in the community of participants so that we reduce selection bias in data collection, thereby reducing prediction bias in competitions; (2) We aim to identify and disseminate best practices in organization of ML competitions that build technological/AI capability in communities that need it; (3) We aim to celebrate innovators and leaders who are using AI in creative ways to improve their communities, so that they can serve as role models and inspire others to do the same.

Workshop Audience

The CiML workshop is targeted at workshop organizers, participants, and anyone with a scientific problem involving machine learning that may be formulated as a challenge. The emphasis of the CiML workshop is on challenge design. Hence it complements nicely the workshop on the NeurIPS 2020 competition track and will help pave the way toward next year's competition program.