Max Tegmark
Last reviewed
May 31, 2026
Sources
16 citations
Review status
Source-backed
Revision
v1 ยท 2,399 words
Improve this article
Add missing citations, update stale details, or suggest a clearer explanation.
Suggest edit
CtrlK
Last reviewed
May 31, 2026
Sources
16 citations
Review status
Source-backed
Revision
v1 ยท 2,399 words
Add missing citations, update stale details, or suggest a clearer explanation.
Max Tegmark is a Swedish-American physicist and artificial intelligence researcher who is a professor of physics at the Massachusetts Institute of Technology (MIT) and the co-founder and president of the Future of Life Institute (FLI). [1][2][3] He spent the first part of his career in cosmology, working on the analysis of the cosmic microwave background (CMB) and the large-scale structure of the universe, and he is also known for the mathematical universe hypothesis. [1][4] His research later moved toward machine learning, the use of physics methods to study and interpret neural networks, and the safety of advanced AI. [2][5]
Through FLI, which he helped establish in 2014, Tegmark has become one of the more visible figures in public debate over existential risk from AI and AI safety. [3][6] In March 2023 the institute published an open letter calling for a six-month pause on the training of AI systems more powerful than GPT-4, a document that drew tens of thousands of signatures and a wide range of criticism. [7][8] He is the author of two popular-science books, "Our Mathematical Universe" (2014) and "Life 3.0: Being Human in the Age of Artificial Intelligence" (2017), the second of which deals with the long-term future of AI. [9][10] In 2023 Time magazine named him to its inaugural list of the 100 most influential people in AI. [11]
Max Erik Tegmark was born on 5 May 1967 in Stockholm, Sweden. [1] His mother is Karin Tegmark and his father is the mathematician Harold S. Shapiro; Tegmark later adopted his mother's surname. [1]
He studied in both Stockholm and the United States. He earned a bachelor's degree in economics from the Stockholm School of Economics in 1989 and a degree in engineering physics from the KTH Royal Institute of Technology in 1990. [1][2] He then moved to California for graduate study at the University of California, Berkeley, where he received a master's degree in physics in 1992 and a PhD in 1994. [1][2] His doctoral work in cosmology was supervised by the astrophysicist Joseph Silk, and his dissertation was titled "Probes of the Early Universe." [1]
After his PhD, Tegmark held a research position at the Max Planck Institute for Physics in Munich, then returned to the United States as a Hubble Fellow and member of the Institute for Advanced Study in Princeton. [1] He joined the faculty of the University of Pennsylvania, where he received tenure in 2003, and moved to MIT in September 2004, where he became a professor of physics. [1][2] At MIT he is affiliated with the Kavli Institute for Astrophysics and Space Research and with the National Science Foundation AI Institute for Artificial Intelligence and Fundamental Interactions (IAIFI). [2]
Much of his cosmology research dealt with extracting physical information from large datasets, including measurements of the cosmic microwave background from the COBE and WMAP satellites and from balloon and ground-based experiments, together with galaxy surveys such as the Sloan Digital Sky Survey and the 2dF Galaxy Redshift Survey. [1][2] He worked on the statistical analysis of temperature fluctuations in the CMB and on the three-dimensional clustering of galaxies, and he contributed to early work, with Daniel Eisenstein and Wayne Hu, on using baryon acoustic oscillations as a cosmological standard ruler. [1] In 2003 he was among the researchers who reported an unexpected alignment in the largest-scale features of the WMAP maps, an anomaly that became known as the "axis of evil." [1] By his own account he has authored more than 300 scientific publications. [3]
Tegmark is also known in the philosophy of physics for the mathematical universe hypothesis, which he set out in a 2008 paper, "The Mathematical Universe," published in the journal Foundations of Physics. [4] The hypothesis holds that physical reality is not merely described by mathematics but is itself a mathematical structure, so that the physical world and an abstract mathematical object are one and the same. [4] He places it within a four-level classification of parallel universes that he has described in articles and in his first book. In this scheme, level one covers regions beyond our cosmic horizon, level two covers regions with different physical constants produced by cosmic inflation, level three is the many-worlds reading of quantum mechanics, and level four is the space of all mathematical structures. [4][9] The idea is speculative and has been debated by physicists and philosophers, with critics arguing that it is difficult to test and resists clear definition. [9]
In March 2014 Tegmark co-founded the Future of Life Institute, a nonprofit organization based in the Boston area. [3][6] Its other founders were the Skype co-founder Jaan Tallinn, the DeepMind researcher Viktoriya Krakovna, the researcher Meia Chita-Tegmark, and the physicist Anthony Aguirre of the University of California, Santa Cruz. [6] Tegmark serves as the institute's president. [3][6]
The institute describes its mission as steering transformative technologies toward benefiting life and away from extreme, large-scale risks. [3][6] Its work spans several areas, including advanced artificial intelligence, biotechnology, nuclear weapons, and climate change, with much of its attention on the long-term effects of artificial general intelligence and superintelligence. [6] In 2015 the organization received a donation of about ten million dollars from Elon Musk to fund a research program on keeping AI beneficial, and in 2021 it received a large cryptocurrency donation from the Ethereum co-founder Vitalik Buterin. [6]
FLI organized two early meetings that helped shape discussion of AI safety. The first was a 2015 conference in Puerto Rico on the opportunities and challenges of AI, which produced an open letter on robust and beneficial AI signed by many researchers. [6] The second was the 2017 Beneficial AI conference at Asilomar in California, where participants drafted the Asilomar AI Principles, an early set of guidelines for AI research and governance. [6][12] Tegmark has also represented the institute in policy settings, including a written statement to a 2023 United States Senate AI Insight Forum. [13]
In March 2023, about a week after the release of GPT-4, the Future of Life Institute published an open letter titled "Pause Giant AI Experiments." [7][8] The letter called on all AI labs to pause immediately, for at least six months, the training of AI systems more powerful than GPT-4, and it asked that the pause be public and verifiable and include all the main developers. [7] It added that if such a pause could not be put in place quickly, governments should step in and impose a moratorium. [7] The letter framed the request around concerns about an out-of-control race to build ever more powerful systems that no one could fully understand or reliably control, and it raised the prospect of automated propaganda, large-scale job loss, and a longer-term loss of human control. [7][8] It also called for shared safety protocols, independent oversight, and more public funding for AI safety research. [7]
The letter attracted more than 30,000 signatures, including those of the deep-learning researcher Yoshua Bengio, the computer scientist Stuart Russell, Elon Musk, the Apple co-founder Steve Wozniak, and the writer Yuval Noah Harari. [8] Tegmark publicly defended it as an attempt to slow a dangerous race and to create space for safety work. [11]
The letter also drew substantial criticism, from several directions. [8] The researcher Timnit Gebru and colleagues argued that it played up speculative, far-future scenarios while drawing attention away from harms that current AI systems already cause, such as bias, labor exploitation, and the concentration of power. [8] Sam Altman, the chief executive of OpenAI, said the letter was missing technical nuance about where a pause would help, and others noted that an early version of the letter implied OpenAI was already training GPT-5, which the company disputed. [8] Bill Gates said he did not think asking one group to pause would solve the problem. [8] From the opposite side, the writer Eliezer Yudkowsky argued that the letter did not go far enough and that a longer, more sweeping halt was warranted. [8] There were also questions in the press about the verification of some signatures in the early hours after publication, since a small number of names were later found to be fake or unconfirmed. [8] No pause occurred, and AI developers continued to train larger systems in the period that followed. [8]
Tegmark has written two books for general readers. "Our Mathematical Universe: My Quest for the Ultimate Nature of Reality," published by Knopf in 2014, lays out his view that reality is fundamentally mathematical and walks through his four-level picture of parallel universes. [9] Reviewers found it ambitious and engaging while questioning the more speculative claims about the multiverse. [9]
His second book, "Life 3.0: Being Human in the Age of Artificial Intelligence," was published by Knopf in 2017 and became a New York Times bestseller. [3][10] The title refers to Tegmark's framing of life in three stages: "Life 1.0" that cannot redesign its own hardware or software, "Life 2.0" such as humans that can redesign its software through learning and culture, and a hypothetical "Life 3.0" that could redesign both its software and its physical form. [10] The book surveys possible futures as AI advances, ranging from beneficial outcomes to scenarios in which humans lose control, and it gives extended attention to the alignment problem, the question of how to ensure that the goals of a highly capable AI stay compatible with human values. [10] Several scientists and publications reviewed it positively, and Tegmark has said the book grew out of his work with the Future of Life Institute. [10]
In the 2010s and 2020s Tegmark shifted a large part of his research toward machine learning, describing his program as using AI for physics and physics for AI. [2] One line of this work concerns mechanistic interpretability, the effort to understand the internal computations of neural networks rather than treating them as black boxes. [5] In a 2023 paper with Wes Gurnee, "Language Models Represent Space and Time," he reported evidence that large language models in the Llama 2 family learn linear internal representations of spatial and temporal information and contain individual neurons that track location and date, which the authors interpreted as a sign that such models build elements of a world model rather than only surface statistics. [14]
Tegmark has also worked on neural network designs intended to be more interpretable and more useful for science. In 2024 he was a co-author, with Ziming Liu and others, of a paper introducing Kolmogorov-Arnold Networks (KANs), an alternative to standard multilayer perceptrons that places learnable functions on the network's connections. [15] The authors argued that KANs can be more accurate and easier to interpret on small science-oriented tasks and can help researchers recover mathematical and physical relationships from data, work they extended in a follow-up paper. [15]
Alongside the research, Tegmark has continued to advocate for caution in the development of the most powerful AI. He has argued, in talks and writing for the Future of Life Institute, for building controllable "tool AI" aimed at specific problems rather than racing toward autonomous general-purpose superintelligence, and he has warned against what he has called an AGI "Manhattan Project." [16] These are his own stated positions. He has often summarized his concern by saying that humanity is closer to building advanced AI than to knowing how to control it, and he treats the alignment problem as the central challenge. [11][16] Other researchers disagree, with some arguing that fears of existential risk are overstated relative to nearer-term harms and practical questions of deployment. [8]
Tegmark was elected a Fellow of the American Physical Society in 2012, and in 2019 he received a gold medal from the Royal Swedish Academy of Engineering Sciences. [1][2] In September 2023 Time magazine included him on its first Time100 list of the most influential people in artificial intelligence, citing his warnings about large-scale risks and his role in the pause letter. [11] Beyond his own books and papers, he is a serial founder of nonprofit organizations, having helped start the Foundational Questions Institute and the Improve the News Foundation in addition to the Future of Life Institute. [3][11]
| Field | Detail |
|---|---|
| Full name | Max Erik Tegmark |
| Born | 5 May 1967, Stockholm, Sweden |
| Nationality | Swedish-American |
| Fields | Physics, cosmology, machine learning, AI safety |
| Education | Stockholm School of Economics (BA, 1989); KTH Royal Institute of Technology (engineering physics, 1990); UC Berkeley (MA 1992, PhD 1994) |
| Doctoral advisor | Joseph Silk |
| Institution | Massachusetts Institute of Technology (professor of physics, since 2004) |
| Known for | CMB and large-scale-structure analysis; mathematical universe hypothesis; Future of Life Institute; 2023 AI pause letter |
| Organizations | Co-founder and president, Future of Life Institute (2014); co-founder, Foundational Questions Institute and Improve the News Foundation |
| Books | "Our Mathematical Universe" (2014); "Life 3.0" (2017) |
| Honors | Fellow, American Physical Society (2012); Gold Medal, Royal Swedish Academy of Engineering Sciences (2019); Time100 AI (2023) |