Roman Yampolskiy
Last reviewed
Jun 8, 2026
Sources
11 citations
Review status
Source-backed
Revision
v1 · 1,483 words
Improve this article
Add missing citations, update stale details, or suggest a clearer explanation.
Suggest edit
CtrlK
Last reviewed
Jun 8, 2026
Sources
11 citations
Review status
Source-backed
Revision
v1 · 1,483 words
Add missing citations, update stale details, or suggest a clearer explanation.
Roman Vladimirovich Yampolskiy (born August 13, 1979) is a Latvian-American computer scientist and a tenured associate professor in the Department of Computer Engineering and Computer Science at the University of Louisville, where he founded and directs the Cyber Security Lab. He is one of the earliest academic researchers in the field of AI safety, a term he is widely credited with coining, and he is among the most prominent voices arguing that a sufficiently advanced artificial superintelligence could not be reliably controlled by humans. [1][2]
Yampolskiy began his career on cybersecurity and behavioral biometrics before turning to the controllability of advanced AI. He has authored or edited several books on the subject, including Artificial Superintelligence: A Futuristic Approach (2015) and AI: Unexplainable, Unpredictable, Uncontrollable (2024), and he argues that the AI control problem may be fundamentally unsolvable. His stark risk estimates, including a widely reported claim that advanced AI poses a 99.9 percent chance of causing human extinction over the next century, have made him a frequent guest on high-profile podcasts hosted by figures such as Lex Fridman and Joe Rogan. [1][3][4]
Yampolskiy was born on August 13, 1979 in Riga, in what was then the Latvian Soviet Socialist Republic of the Soviet Union, and he later emigrated to the United States. [1]
He earned an associate's degree in computer science from Monroe Community College in 2000 and a combined bachelor's and master's degree in computer science from the Rochester Institute of Technology in 2004. He completed a PhD in computer science and engineering at the University at Buffalo in 2008, supported by a four-year National Science Foundation IGERT fellowship. His doctoral research, supervised by the biometrics specialist Venu Govindaraju, focused on behavioral biometrics: the use of distinctive patterns in how an agent behaves, rather than physical traits such as fingerprints, to verify identity. His dissertation, "Intrusion Detection Using Spatial Information and Behavioral Biometrics," applied these techniques to computer security, including an early line of work on detecting automated "poker bots" by building a behavioral profile of each online player and flagging significant deviations to game administrators as possible security breaches. [1][5][6]
After his doctorate, Yampolskiy held an affiliate academic appointment at University College London's Centre for Advanced Spatial Analysis and joined the University of Louisville as an assistant professor in 2008. He was promoted to associate professor with tenure in 2014. [1][5]
At the University of Louisville, Yampolskiy founded the Cyber Security Lab and has served as its director since around 2012. His early research spanned pattern recognition, biometrics, computer security, and the security of online games, and he has published more than 200 peer-reviewed papers over his career. A 2023 analysis of citation data placed him among the top 2 percent of cited scientists worldwide. [1][2][5]
Over time his attention shifted toward the safety of advanced artificial intelligence. In a 2011 paper titled "Artificial Intelligence Safety Engineering: Why Machine Ethics Is a Wrong Approach," first presented at the Philosophy and Theory of Artificial Intelligence conference in Thessaloniki, Greece, he argued that efforts to give machines moral reasoning or rights were misguided and that researchers should instead treat the challenge as one of safety engineering, building systems that can be proven safe even under recursive self-improvement. He is widely credited with helping to establish and popularize the term "AI safety" around this period. [1][7]
In 2015 Yampolskiy proposed "intellectology," which he described as a new field founded to analyze the forms and limits of intelligence, encompassing both human and artificial minds and the broader "space of possible mind designs." This framing of intelligence as a space to be mapped underpins much of his later argument that superintelligent systems would be alien and difficult to reason about. [1][2]
Yampolskiy is best known for the argument that a superintelligent AI, a system far more capable than humans across essentially all domains, could not be permanently and reliably controlled. He frames this as the AI control problem and contends that there is no evidence a solution exists. To illustrate the asymmetry between human overseers and a vastly more capable system, he has said that "imagining humans can control superintelligent AI is a little like imagining that an ant can control the outcome of an NFL football game being played around it." [3][4]
His 2024 book AI: Unexplainable, Unpredictable, Uncontrollable develops three interlocking claims: that the decisions of advanced AI systems are often unexplainable, that their behavior is unpredictable, and that, as a consequence, such systems are ultimately uncontrollable. Because the ability to verify that a system is safe cannot keep pace with its growing capability, he argues, developers cannot guarantee that an advanced system will remain aligned with human intentions. This connects his work to the broader research programs of AI alignment and existential risk from AI. [4][9]
Yampolskiy distinguishes among several categories of bad outcome. Beyond conventional existential risk, in which humanity is destroyed, he highlights "suffering risks," outcomes judged worse than extinction, and what he calls "ikigai risk," a future in which AI strips human life of meaning and purpose even if people physically survive. [4]
He has offered some of the most extreme public probability estimates in the debate. Appearing on the Lex Fridman Podcast in June 2024, he put the chance that AI leads to human extinction at "99.9 percent within the next hundred years," a figure that was widely reported in the press. He stresses that such estimates are inherently uncertain and depend on assumptions about how advanced systems are built and deployed. By contrast, on The Joe Rogan Experience in July 2025 he reported that many leading figures inside the AI industry privately put the probability of human extinction from AI at roughly 20 to 30 percent, far below his own estimate but, in his view, still unacceptably high. [3][4][10]
Because he doubts the control problem can be solved, Yampolskiy advocates pausing or indefinitely delaying the development of artificial general intelligence and superintelligence, and he has supported research into "boxing," or confining, advanced systems so their influence on the world is limited. In March 2023 he was among the signatories of the Future of Life Institute open letter "Pause Giant AI Experiments," which called for a moratorium on training the most powerful AI models. [1][3]
Yampolskiy has written or edited several books on artificial intelligence and its risks, most published by CRC Press and its Chapman and Hall imprint.
| Year | Title | Role | Notes |
|---|---|---|---|
| 2015 | Artificial Superintelligence: A Futuristic Approach | Author | His first monograph on superintelligence and the control problem [8] |
| 2018 | Artificial Intelligence Safety and Security | Editor | First edited volume devoted to safe and secure advanced AI, with 28 chapters [1] |
| 2024 | AI: Unexplainable, Unpredictable, Uncontrollable | Author | Argues that advanced AI cannot be fully explained, predicted, or controlled [9] |
| 2025 | Considerations on the AI Endgame: Ethics, Risks and Computational Frameworks | Co-author, with Soenke Ziesche | Examines the ethical and computational dimensions of advanced AI [1] |
Yampolskiy is a research advisor to the Machine Intelligence Research Institute and an AI safety fellow at the Foresight Institute, and he has been associated with the Global Catastrophic Risk Institute. He is a senior member of the Institute of Electrical and Electronics Engineers (IEEE). [1][2]
His blunt warnings about advanced AI have given him a substantial public profile. His June 2024 conversation with Lex Fridman and his July 2025 appearance on The Joe Rogan Experience, episode 2345, introduced his arguments to large general audiences, and he is a frequent commentator in technology and mainstream media. [3][10]
He has received numerous honors over his career, including recognition on lists of leading figures in artificial intelligence and, in 2023, a ranking among the top 2 percent of cited scientists worldwide in a Stanford University analysis. In 2025 he received the Lifeboat Foundation's Guardian Award. As of 2026 he remains on the faculty of the University of Louisville and continues to write and speak about the risks of advanced AI. [2][5][11]