Ilya Sutskever is a computer scientist specializing in machine learning and artificial intelligence who serves as co-founder and Chief Scientist of Safe Superintelligence Inc. His career has centered on deep learning research, contributing to foundational advances in neural networks, large language models, reasoning systems, and AI safety. ^[1]

Education

Sutskever graduated from the University of Toronto with a Bachelor’s in Mathematics in 2005 and later a PhD in Computer Science in 2012. ^[2]

Career

Sutskever studied mathematics and computer science at the University of Toronto, where he completed his doctoral research under Geoffrey Hinton. During this period, he collaborated with Hinton and Alex Krizhevsky to develop AlexNet, a convolutional neural network that demonstrated significant improvements in image recognition and helped accelerate the adoption of deep learning. Following his PhD, he completed a brief postdoctoral appointment at Stanford University in Andrew Ng's research group before co-founding DNNResearch in late 2012, a company established to advance deep learning research.

After Google acquired DNNResearch in 2013, Sutskever joined Google Brain as a research scientist. There, he worked on several influential machine learning projects, including the development of the sequence-to-sequence learning architecture with Oriol Vinyals and Quoc Le, contributed to TensorFlow, and co-authored research associated with AlphaGo. His work during this period helped establish techniques that became widely used across natural language processing and artificial intelligence.

In 2015, Sutskever co-founded OpenAI, initially serving as Research Director before becoming Chief Scientist in 2018. He led research programs that contributed to the development of GPT models, ChatGPT, reasoning models, CLIP, DALL·E, and other large-scale AI systems, while helping shape the organization's long-term research strategy. He also co-led OpenAI's Superalignment initiative, which focused on research into the alignment and safety of advanced AI systems. In 2023, he briefly served on OpenAI's board of directors during the leadership changes involving CEO Sam Altman before stepping down from the board later that year.

In June 2024, Sutskever left OpenAI and co-founded Safe Superintelligence Inc., where he serves as Chief Scientist. The company focuses exclusively on research and development of advanced AI systems, with an emphasis on safety, aligning with Sutskever's long-standing research interests in artificial intelligence and alignment. ^{[2] [3]}

Interviews

Age of Research

On the Dwarkesh Podcast in November 2025, Sutskever reflected on the remarkable reality of AI development, emphasizing how society quickly adapted to the rapid scaling of AI capabilities that seemed like science fiction just a few years prior. He discussed the disconnect between models' impressive evaluation performance and their actual economic and real-world impact, noting that models often repeat mistakes despite achieving superhuman performance on specific tasks. He attributed this partly to the narrow focus imposed by reinforcement learning training and to limitations in generalization, highlighting human-like learning efficiency, robustness, and the mysterious nature of evolved social desires and emotions. Sutskever argued that the current era was returning to an "age of research" rather than mere scaling, emphasizing the importance of understanding fundamental principles such as generalization and continual learning in building safe and aligned superintelligent systems. He suggested that progress hinges on exploring new approaches such as incremental deployment, the role of value functions, and fostering diversity among AI agents, rather than solely increasing scale. He expressed optimism that multiple organizations, including SSI, could develop strategies emphasizing alignment and safety by focusing on understanding human-like learning and value representations. Finally, he underscored the significance of research driven by aesthetics, biological inspiration, and simplicity, believing that combining these principles with empirical progress will guide the future of powerful AI systems and their safe deployment. ^[6]

Presentations

Journey Towards AGI

In a TED Talk in November 2023, Sutskever discussed the rapid advancements in artificial intelligence, emphasizing that AI, essentially digital brains within computers, had the potential to surpass human intelligence, leading to the development of artificial general intelligence (AGI). He shared his personal motivations for entering the field, including curiosity about consciousness and the desire to understand and replicate intelligence. Sutskever explained that although current AI systems are less intelligent than humans, they will eventually become as capable or more so, profoundly impacting every aspect of life, such as healthcare, where an AGI could revolutionize medical treatment by offering exhaustive knowledge and constant availability. He highlighted the transformative nature of AGI, acknowledging both its immense benefits and associated risks, including the possibility of rapid self-improvement and unintended consequences. To address these concerns, he noted efforts by AI organizations such as OpenAI to ensure safety and foster collaboration among competitors. Reflecting on recent rapid progress and heightened awareness, Sutskever expressed his optimism, believing that collective cooperation driven by self-interest would help humanity navigate the challenges posed by AGI’s emergence. ^[5]