In the coming months, we will have a searchable resource library to support your learning. Until then, here are a few items to get you started.
AI and disability, small minorities and outliers
The article discusses challenges persons with disabilities face with current AI systems and approaches that need to be adopted in AI development to ensure fairness.
A report capturing themes and discussions from a workshop at AI NOW Institute at New York University, the NYU Center for disability studies workshop.
Provides specific examples of how certain AI technologies (facial, body, object, text, speech recognition; speech and text generation and analysis; biases can impact fairness for persons with disabilities.
Work for people with disabilities in data science
A joint publication by Fundación ONCE and the ILO Global Business and Disability Network, developed within the framework of Disability Hub Europe, a project led by Fundación ONCE and co-funded by the European Social Fund.
AI ethics and policy
To provide needed insight into the current state of practice in the industry, the article presents survey data from 211 software companies. The data provides some context for this special issue by helping us understand where we currently are as an industry in terms of AI ethics. For practitioners, the data can also serve as a way to benchmark where your organization stands.
An overview of ethical issues in AI including privacy, transparency, explainability and bias, and what needs to be done to address these issues.
Technical information (for AI experts)
Authors designed, applied and evaluated a social participatory framework (WeBuildAI) for engaging community stakeholders to create a decision making algorithm for their needs.
Youtube session from Google I/O ’19 explaining ML to those with coding experience. Discusses image classification problem and use of Tensor flow.
From Google I/O, YouTube video on fairness lessons learned through their products and research and describes techniques and resources to support fairness.
From Google PAIR, Google’s explanation on how fairness is treated using Tensorflow
This article argues that traditional statistical methods were developed for small data sets and not suitable for current large and complex data sets.