Scholar
Dmitrii Krasheninnikov
Google Scholar ID: BIQflKQAAAAJ
University of Cambridge
Citations & Impact
Citations
1,684
H-index
8
i10-index
8
Publications
15
Co-authors
0
Publications
Resume (English only)
Academic Achievements
- Coined the term “out-of-context learning”; paper 'Training-order recency is linearly encoded in language model activations' was a runner-up for best paper at the MemFM workshop @ ICML 2025; contributed to several papers published at top conferences like NeurIPS and ICML.
Research Experience
- Works on AI safety at Cambridge; worked with UC Berkeley’s Center for Human-Compatible AI; conducted research on deep RL and robotics at Sony AI Zurich.
Education
- PhD student at Cambridge, supervised by David Krueger and Rich Turner; MSc in AI from the University of Amsterdam.
Background
- Research interests include AI safety, interpretability of deep learning, control, and security. He believes that in the next decade, we may develop AI systems capable of doing everything humans can do using computers, and emphasizes the need to ensure that deploying such systems won’t lead humanity to permanently lose control over our future.
Miscellany
- Likes nature and AI safety
Co-authors