In this two-part course, you learned the basics of Crowdsourcing, an important method that can be used to gather data for building AI systems by leveraging human intelligence at scale for data creation, enrichment, and interpretation.

In the first part, we discussed the intuition behind Crowdsourcing and how this paradigm can be used to collect high-quality data for AI solutions. We zoomed into the hurdles of Crowdsourcing and focused on quality control for Crowdsourcing.

In this space, we discussed both the mechanisms that enable researchers and practitioners to carry out quality control as well as the human factors that affect the overall quality of the crowdsourced data.

We then built on top of the knowledge acquired in the first part to shift our focus to using Crowdsourcing as a means to explain and diagnose Machine Learning models and pipelines.

In the second part, we started by explaining the role Crowdsourcing has in advanced Machine Learning.

Motivated by the large amount of data AI systems commonly need, we discussed the Active Learning framework and how this eases labelling efforts.

Finally, we concluded the second part by overviewing progress at the intersection of Crowdsourcing and Interpretability and debugging of Machine Learning models. Throughout the course we offered additional content, readings and videos. We hope these actively helped you throughout your learning experience!