The Electronic Frontier Foundation (EFF) started a document containing progress artificial intelligence (AI) research on multiple tasks.
The goal of the document is to be the place for people to find progress on difficult tasks. Currently, many tasks don't have the metrics, datasets, and benchmarks to keep track of them. To bootstrap the overview of the state of the art in AI, they went through many blog posts and websites. The EFF made a Jupyter Notebook to which researchers and developers can contribute on Github. They encourage others to create pull requests for additional datasets and include the speed and efficiency of training methods.
The reason behind the overview is that the EFF wants to know the state of AI for political reasons. Their goal is to learn what types of AI they have to engage with on legal, political, and technical safety fronts. Through this overview journalists, policy makers, and technology users can understand how far the field of machine learning and AI is at the moment. Because algorithms can be designed to violate privacy, threaten safety, and perpetuate inequality and injustice, people must be able to anticipate these risks before they arise.
People can also contribute by adding visualizations to their data. Progress is continuously updated. Although the project started as an EFF project, the goal is to grow it to be a self-sustaining community effort.
