Alexander Ku
I'm a research scientist on the Foundational Research team at Google DeepMind and a PhD candidate in the Department of Psychology at Princeton University.
My research focuses on how AI systems allocate limited cognitive resources -- such as memory (e.g., context, retrieval, weights) and compute (e.g., reasoning, planning, search, verification, communication) -- in both single- and multi-agent settings, with a particular focus on how they can become more resource-efficient with experience.
Prior to this, I've worked on text-to-image generation, vision-language models, embodied agents, and applications of machine learning to genomics.
Email / Google Scholar / CV / GitHub / Twitter
Selected papers
Recent or representative papers:
- Ku, A., Campbell, D., Bai, X., Geng, J., Liu, R., Marjieh, R., ... & Griffiths, T. (2026). Levels of analysis for large language models. Transactions of the Royal Society A: Mathematical, Physical and Engineering Sciences, 384(2320). (pdf)
- Mieczkowski, E., Ku, A., Eisape, T., Arumugam, D., Matters, J., Collins, K. M., ... & Griffiths, T. L. (2026). Improving the Efficiency of Language Agent Teams with Adaptive Task Graphs. preprint arXiv:2605.06320. (pdf)
- Ku, A., Griffiths, T. L., & Chan, S. C. (2026). An evolutionary perspective on modes of learning in Transformers. The Fourteenth International Conference on Learning Representations (ICLR). (pdf)