What is Machine Learning?
Machine Learning (ML), commonly categorized as a subfield of Artificial Intelligence (AI), is a field of study concerning how to automatically extract meaningful information from data with statistical algorithms. ML’s driving principle is that an algorithm can use statistical patterns within data to make accurate predictions about new data that it hasn’t seen before. While ML stems from statistical modeling, it differs in that the extraction and extrapolation process is done automatically and can also be conducted without human guidance. Once the algorithm is trained on sample data, researchers and software engineers can apply the algorithm to new, larger streams of data. ML is already an integral component of many deployed commercial applications, such as social media feed ranking, financial market prediction, and medical image diagnoses, as well as public applications, such as risk assessments in the justice system. In addition, ML will be foundational in the future of various other emerging technologies, such as autonomous vehicles (AVs) and next-generation cybersecurity.
Despite ML’s widespread deployment, domestic regulation remains sparse. While an Artificial Intelligence in Government Act was recently introduced in the Senate, the most comprehensive regulation stems from the European Union’s General Data Protection Regulation (GDPR), which restricts the areas where autonomous decisions are legal and gives individuals greater control over their data. Without clear guidance and regulation, the expansion of ML raises significant policy challenges, such as transparency and explainability, accountability, and fairness.
Download the Full Document:
Amy Robinson and Ariel Herbert-Voss. Ed. Bogdan Belei. “Technology Factsheet Series: Machine Learning” Paper, Belfer Center for Science and International Affairs, Harvard Kennedy School, June 2019.