To follow its mission to allow anyone, anywhere, to benefit from AI for the common good; AI Commons fosters collaboration and initiates AI-based problem-solving projects. Every initiative and project AI Commons incubates is based on the principles of collaboration and sharing.
AI Commons is an open framework and works with other organizations, initiatives and individuals that align with its mission. AI Commons intends to leverage/use existing research, guidelines, and work already done by other groups that will match its mission.
AI Commons organized a Health and Well-being Hackathon in partnership with Data Science Nigeria in June 2020. The hackathon brought together 100 people for an online ideation workshop. Participants included problem owners such people who personally experienced health challenges in the region, as well as health workers, students, academic researchers, solution developers, and technical experts.
Ideation workshop started with raising awareness about local health challenges with existing solutions. The community of problem owners and problem solvers then explored ways to improve existing solutions by evaluating them and identified new ones. Organizers and participants co-built an open collaboration model and methodology that can be used in other contexts and improved upon.
LIFE360 is an accessible and decentralized framework of collaboration for AI-based-problem-solving for all.
AI Commons Repository is a common repository of trained ML/AI models along with their training data sets that have been used for AI For Good solutions to serve as a resource for all communities of practice to advance applications of ML/AI techniques for AI For Good purposes.
Connecting problem owners to problem solvers is important to identify how real challenges on the ground that can be addressed using AI. Living and learning labs around the world acting as AI community hubs help matching problems solvers with problem owners, identify data sources and help build solutions that can be immediately used locally. Here are some of the active AI Community Hubs
AI Commons developed an extensive template for AI solutions. The solution template aims to increase the accessibility, reproducibility, contextualization, and enhancement of Artificial Intelligence solutions locally and globally.
In this challenge, participants develop an algorithm that can generate a character profile when provided the name of a prominent female in history, science, technology, or culture including folklore and myth. The result of the challenge is contributed to AI Commons to be used globally and incentivize for more diverse and women focused inspiration on innovation.
The Global Data Commons (GDC) aims at proposing models to support the creation of data commons on a global basis, and providing mechanisms for rules-based access to data and help leveraging the revolution in advanced analytics and Artificial Intelligence to support the achievement of the UN Sustainable Development Goals (SDGs).
To help all problem solvers work on real cases with problem owners, model AI sandbox are open “Safe environments” to be used for collaboration and solution evaluation. These sandboxes incorporate AI/ML algorithms, training data sets, and cloud and compute resources, and help anyone to start evaluating real life solution evaluation.
GenderNet is a project initiated by AI Commons. Machine learning researchers need benchmark datasets to train their algorithms. Biased and non-representative datasets lead to harmful AI solutions that systematically undermine underrepresented groups such as women and LGBTQIA++. To close the gender data gap, AI Commons with ITU launched a call for ethically sourced, diverse, and balanced gender dataset that will help make biases visible. GenderNet is inspired by the infamous ImageNet. Balanced gender representation in datasets is not a silver bullet solution to eliminate bias, however it’s a good starting point to craft AI solutions that support social good.
AI Commons launched the Global Data Pledge in partnership with ITU’s Global Initiative on AI and Data Commons. The Global Data Pledge functions on the basis of collective action to make data available around ‘trigger events’ such as Covid-19. The goal is to make data available as a public resource especially in the face of global events and challenges. The Global Data Pledge is open to all interested parties. It is collaborative and transparent and focuses on impact and scalability. Fill in this form to join us to make data available when it matters.
ImpactNet is an AI for Impact benchmark database. This initiative will produce model datasets and model AI solutions. Datasets will allow more researchers to use their models on new areas that will have an impact on helping achieve AI for Good. AI solutions will allow problem solvers to adopt validated AI for good solutions to scale those solutions in areas of SDGs to accelerate impact.