Ilya Sutskever (b. 1985, Nizhny Novgorod, Russia) is a computer scientist known for his work on neural networks, deep learning, and artificial intelligence systems. He is one of the co-founders of OpenAI and served as its chief scientist until 2024. As of 2024, he is the founder and leader of Safe Superintelligence Inc., a new artificial intelligence venture.

Early Life and Education

Sutskever was born in Nizhny Novgorod during the Soviet era and moved with his family to Jerusalem at the age of five. During high school, he took computer science courses through the Open University of Israel. After his family immigrated to Canada, he completed his undergraduate degree in mathematics at the University of Toronto. He then earned his PhD in computer science at the same university under the supervision of Geoffrey Hinton, a renowned researcher in artificial intelligence. During his studies, he received the Google Graduate Fellowship from 2010 to 2012 and the Ontario Graduate Scholarship from 2006 to 2009.

Academic and Professional Career

After completing his PhD in 2012, Sutskever briefly served as a postdoctoral researcher in Andrew Ng’s group at Stanford University. That same year, he co-founded DNNResearch, an artificial learning startup, with Hinton. The company was acquired by Google in 2013, and Sutskever joined the Google Brain team.

Scientific Contributions

Sutskever’s contributions to artificial intelligence research center on various model architectures and learning techniques:

• AlexNet (2012): Developed with Geoffrey Hinton and Alex Krizhevsky, this deep convolutional neural network model demonstrated the effectiveness of deep learning in image processing by achieving groundbreaking success in the 2012 ImageNet competition.
• Sequence-to-Sequence Learning (2014): Developed with Oriol Vinyals and Quoc Le, this technique enabled the execution of numerous natural language processing tasks, particularly machine translation.
• TensorFlow (2015): Played an active role in the development of TensorFlow, an open-source machine learning library, while working at Google Brain.
• AlphaGo (2016): Contributed to the development of the system that combined neural networks with tree search techniques.
• Deep Reinforcement Learning: Advanced this approach within OpenAI to model decision-making processes in robotics and game environments.
• GPT Series: Provided technical leadership in the development of large language models such as GPT-2, GPT-3, and GPT-4 at OpenAI.
• DALL·E and CLIP: Played a role in developing models that jointly process visual and linguistic data.
• Codex: Contributed to the training of this system that converts natural language inputs into code outputs for programming tasks.

OpenAI and Superintelligence Research

In 2015, Sutskever became a co-founder of OpenAI, led by Elon Musk and Sam Altman. He served as chief scientist and worked within OpenAI’s mission to develop artificial intelligence for public benefit. He played an active role in developing projects such as GPT systems, DALL·E, and Codex.

Sutskever initiated the Superalignment project, focused on ensuring the safe development of artificial superintelligence. This initiative aims to develop highly capable AI systems that are aligned with human values and controllable.

Leadership Crisis and Departure from OpenAI

In November 2023, Sutskever played a role in the removal of OpenAI’s CEO Sam Altman. The decision sparked widespread controversy, and Altman was soon reinstated as CEO. In a public statement following the events, Sutskever expressed regret over his involvement and announced his departure from OpenAI a few months later.

Safe Superintelligence Inc.

In May 2024, Sutskever announced the founding of Safe Superintelligence Inc., a new artificial intelligence company, alongside Daniel Levy. The organization operates as an independent research institute focused on the safe construction of superintelligence.

Awards and Recognition

Throughout his career, Sutskever has received numerous scientific awards and fellowships:

• Google Graduate Fellowship (2010–2012)
• Ontario Graduate Scholarship (2006–2009)
• University of Toronto “Innovator of the Year” Award (2014)
• MIT Technology Review “35 Under 35” Innovators List (2015)
• Fellow of the Royal Society (2022)
• TIME Magazine’s 100 Most Influential People (AI Category – 2023, 2024)

Notable Academic Publications

• Krizhevsky, A., Sutskever, I., & Hinton, G. E. (2012). ImageNet Classification with Deep Convolutional Neural Networks. Advances in Neural Information Processing Systems, 25.
• Srivastava, N., Hinton, G., Krizhevsky, A., Sutskever, I., & Salakhutdinov, R. (2014). Dropout: A Simple Way to Prevent Neural Networks from Overfitting. Journal of Machine Learning Research, 15(1), 1929–1958.
• Sutskever, I., Vinyals, O., & Le, Q. V. (2014). Sequence to Sequence Learning with Neural Networks. Advances in Neural Information Processing Systems, 27.
• Brown, T. B., Mann, B., Ryder, N., Subbiah, M., Kaplan, J. D., Dhariwal, P., … Sutskever, I. (2020). Language Models are Few-Shot Learners. Advances in Neural Information Processing Systems, 33, 1877–1901.
• Mikolov, T., Sutskever, I., Chen, K., Corrado, G. S., & Dean, J. (2013). Distributed Representations of Words and Phrases and their Compositionality. Advances in Neural Information Processing Systems, 26.
• Abadi, M., Agarwal, A., Barham, P., Brevdo, E., Chen, Z., Citro, C., … Sutskever, I. (2016). TensorFlow: Large-Scale Machine Learning on Heterogeneous Distributed Systems. arXiv:1603.04467.
• Radford, A., Kim, J. W., Hallacy, C., Ramesh, A., Goh, G., Agarwal, S., … Sutskever, I. (2021). Learning Transferable Visual Models From Natural Language Supervision. International Conference on Machine Learning, 8748–8763.
• Ramesh, A., Pavlov, M., Goh, G., Gray, S., Voss, C., Radford, A., … Sutskever, I. (2021). Zero-shot Text-to-Image Generation. International Conference on Machine Learning, 8821–8831.
• Szegedy, C., Zaremba, W., Sutskever, I., Bruna, J., Erhan, D., Goodfellow, I., & Fergus, R. (2013). Intriguing Properties of Neural Networks. arXiv:1312.6199.