New York Site views —Page views —
← All series

Directions in AI research

Explore research on retrieval, agents, reasoning, evaluation and multimodal systems.

Reading order

  1. Retrieval-augmented generation: from looking up information to verifiable reasoning
  2. Tool learning: from Toolformer to reinforcement-learned tool policies
  3. Long-context language models: fitting information versus using it
  4. Evaluating LLMs: dynamic tasks and contamination-resistant benchmarks
  5. Prompt injection and agent security: moving toward permission-aware architectures
  6. When language models become judges: a review of reliability research
  7. Multimodal agents: from reading screens to acting in the world
  8. AI research agents: from literature review to experimental design
  9. Agent memory: from vector stores to manageable long-term memory
  10. Test-time scaling: search, verifiers, reflection and self-consistency
  11. Efficient LLM serving: KV caching, speculative decoding, PagedAttention and quantization
  12. Synthetic data, distillation and the management of post-training data
  13. Small language models, on-device AI and edge deployment
  14. Software engineering agents: generation, debugging, evaluation and repository tasks
  15. AI for science: from research assistants to closed-loop experiments
  16. Language models for mathematical reasoning: theorem proving, verification and Olympiad problems
  17. Language models for code: program synthesis, automated debugging and software evaluation
  18. World models and embodied AI: from video prediction to interactive environments
  19. Neuro-symbolic AI: symbolic reasoning, program induction and interpretable structures
  20. Mechanistic interpretability: from circuits to representation analysis
  21. Alignment science and controllable generation: preferences, constitutional training and verification