From: lexfridman
In recent years, machine learning has emerged as a key component in the evolution of modern engineering and decision-making processes. Its potential extends far beyond what many initially anticipated, creating new paradigms in how systems are built and how decisions are made. Michael I. Jordan, a leading figure in the realm of machine learning, provides valuable insights into this transformative role.
Machine Learning as a Proto-Form of Engineering
Machine learning is often described as a “proto-form of engineering” that is distinct from traditional engineering disciplines like chemical or electrical engineering [00:05:00]. Jordan explains that machine learning is built on statistical and computational principles and aims to develop systems that bring value to people by using human data and integrating human decisions. Unlike conventional engineering, which draws upon well-understood theoretical foundations, machine learning as an engineering discipline is still emerging and continues to evolve at a rapid pace.
Machine Learning Challenges
The engineering side of machine learning remains ad hoc, suggesting significant room for development and standardization in practices and methodologies [00:05:47].
Expanding the Scope of Artificial Intelligence
In a blog post appropriately titled “Artificial Intelligence: The Revolution Hasn’t Happened Yet,” Jordan advocates for broadening the scope of the field of artificial intelligence. He argues that the field should not only focus on engineering algorithms and robots but also on understanding and empowering human beings across different levels of abstraction—from individuals to entire civilizations [00:01:04].
This broad view underscores the need to develop a new branch of engineering that integrates human-centric considerations into the design and implementation of AI systems.
Decision Making in a Messy World
The application of machine learning extends into decision-making, particularly in how these systems help make decisions at scale and in distributed environments with inherent uncertainty [00:23:02].
"Prediction plus decision making is everything, but both are equally important." — Michael I. Jordan [00:22:25]
Machine learning enables better decision-making by integrating aspects like risk evaluation and economic context, areas often ignored in simplistic predictive models. As Jordan emphasizes, the goal is not just prediction but taking actions that consider real-world constraints and influence.
Creating New Markets and Opportunities
Machine learning’s role in engineering also includes creating new markets. As highlighted by the music industry example provided by Jordan, machine learning can facilitate connections between producers and consumers, leading to new forms of economic activities and opportunities for individuals and small businesses that were previously inaccessible [00:26:09].
The Future: Beyond Traditional AI
The scope of machine learning is expanding beyond traditional artificial intelligence paradigms. While the concept of artificial intelligence excites imaginations, it traditionally frames the issue too narrowly. Jordan suggests a broader discipline is emerging—a human-centric engineering discipline—capable of leveraging the capabilities of machine learning in diverse ways to address complex societal needs [01:45:14].
Engineering a Future
Jordan envisions this emerging discipline as one that is deeply integrated with human life, characterized by robust, adaptive, and economically viable systems [01:31:00].
In conclusion, the role of machine learning in engineering and decision-making is not limited to advancing technical capabilities. It encompasses creating value through new market structures, improving decision-making processes under uncertainty, and eventually forming a new branch of engineering that is fully aware of human needs and societal dynamics. This transformation will require a change in how the field is perceived and how success is defined, moving from the narrow aspirations of traditional AI to a broader, more impactful vision for machine learning in human life.