The AI Knobs, Tips and Tricks to Jumpstart AI projects

Presented by Anusua Trivedi, Sr Data Scientist Lead, AI for Good at Microsoft

AI serves the purpose of enabling human beings in making better decisions. In this session, we talk about how the actions of AI are the result of the human inputs going into its programming. We talk about how an AI’s bias is not its own, but the human bias with which it has been programmed. We emphasize the choice of the right metric and the type of data used for testing and training to avoid such bias. We discuss the need to understand the dependence between the data used and the models employed and optimize only areas that matter. We discuss how to focus on feature engineering and be thoughtful about the ethics of ML applications. Other issues such as the need for regulations and other considerations within it that require deliberation are also touched upon.