Scholar
Alexander Ku
Google Scholar ID: Lh_ZqdcAAAAJ
Research Scientist, Google DeepMind, Princeton University
Citations & Impact
Citations
7,101
H-index
15
i10-index
17
Publications
20
Co-authors
13
list available
Publications
Resume (English only)
Academic Achievements
- Selected papers:
- - Levels of analysis for large language models
- - Predictability shapes adaptation: An evolutionary perspective on modes of learning in transformers
- - On the generalization of language models from in-context learning and finetuning: A controlled study
Research Experience
- Works as a research scientist at Google DeepMind, focusing on how the mind combines familiar parts to solve unfamiliar problems (compositionality), how those parts are represented and processed, and how these representations adapt to make solving familiar problems more efficient over time (automaticity). Also explores how insights from cognitive science can help evaluate the capabilities and limitations of frontier models.
Education
- PhD student at Princeton University, advised by Tom Griffiths, Jon Cohen, and Mike Mozer.
Background
- PhD candidate at Princeton and a research scientist at Google DeepMind. Research interests include continual learning, meta-learning, and resource-rationality. Focuses on the computational principles underlying the flexibility of human cognition and whether these principles extend to large language models.
Miscellany
- Contact information includes Email, Google Scholar, GitHub, Twitter, and CV.
