Stephen Wolfram discusses the current state of AI in scientific discovery, emphasizing that no AI has yet made a significant breakthrough in decoding the universe's fundamental laws. He highlights the importance of paradigm shifts, the role of computational experiments in emerging fields like reology, and the necessity of deep knowledge in established sciences. Wolfram also reflects on his own experiences and the potential of computation to reveal complex patterns from simple rules.
Stephen Wolfram, a renowned scientist and CEO of Wolfram Research, shares his insights on the current capabilities of artificial intelligence (AI) in scientific discovery, the nature of computation, and the future of understanding the universe's fundamental laws.
Wolfram begins by expressing a candid view: no AI has truly impressed him yet in terms of making a big breakthrough in understanding the "machine code of the universe." He explains that instead of relying solely on equations, using programs to think about the world allows us to derive the laws of physics. This approach, he notes, was something he previously thought impossible.
He emphasizes that to make significant progress in science, one must have a paradigmatically different approach. The recent merger of computational technology with large language models (LLMs) adds a layer of humanization to the process, but it is not yet sufficient to revolutionize scientific discovery.
Wolfram reflects on the motivation behind trying to develop a theory of everything. He points out that there are problems and questions that scientists almost solved a century ago but got stuck on. Now, with new computational tools and methodologies, these problems can potentially be "unstuck," which he describes as magical.
During the interview at the London Institute for Mathematical Sciences, in the historic study of Michael Faraday, Wolfram discusses the importance of being part of the scientific establishment.
For emerging fields where little is known, one does not need extensive prior knowledge to contribute. However, in well-established fields like physics, deep knowledge is essential to make meaningful progress. He notes that many people write theories based on high school physics, but these often fail because modern physics is complex and requires years of study.
Wolfram introduces the concept of "reology," the study of simple computational rules and their behaviors, such as cellular automata and Turing machines. This is a young and wide-open field where many can contribute without needing extensive formal education.
He predicts that in the future, there will be degrees and departments dedicated to reology. Once the foundational knowledge accumulates, contributing at the frontiers will require deep understanding.
Wolfram explains a common pattern in the history of science:
He cites his own experience in the late 1970s and mid-1970s when particle physics was undergoing such a phase of rapid growth due to new methodologies. This allowed even young scientists to make significant contributions.
Wolfram defines computation abstractly as having a set of rules and working out their consequences. For example, a simple rule might determine the color of cells in a line based on neighboring cells. Running such rules millions of times can produce complex patterns.
When he began exploring this in the early 1980s, he expected simple rules to produce simple patterns. However, he discovered that simple rules can generate highly complex and unexpected behaviors.
Stephen Wolfram's reflections highlight the nuanced relationship between AI, computation, and scientific discovery. While AI has not yet made groundbreaking discoveries in understanding the universe's fundamental laws, the integration of computational methods and human insight holds promise. Emerging fields like reology offer new frontiers for exploration, and methodological advances continue to drive scientific progress. Wolfram's work underscores the complexity of the universe and the potential of computation to unlock its secrets.
