Skip to content
AI.info

AI.info

Generative AI

Generation as a system rather than a magic box: distributions, tokens, grounding, and the controls around the model call.

  1. Generative AI as a System, Not a Magic Box
  2. From Probability Distributions to Generated Artifacts
  3. Tokens, Tokenizers, and the Boundaries a Model Sees
  4. Embeddings and Representation Spaces
  5. Encoder, Decoder, and Encoder–Decoder Model Families
  6. Inside a Transformer Block: An Operational Anatomy
  7. Autoregressive Language Modeling and the Next-Token Objective
  8. Pretraining Data, Objectives, and the Behavior They Create
  9. Scaling Laws, Data Budgets, and Compute Allocation
  10. Benchmarks, Contamination, and Capability Claims
  11. Post-Training as Behavior Shaping
  12. Instruction Tuning and Supervised Fine-Tuning
  13. Preference Data, Rubrics, and Reward Models
  14. RLHF, DPO, and Preference Optimization Tradeoffs
  15. Decoding, Search, and Sampling
  16. Prompting as Interface Engineering
  17. Few-Shot Demonstrations and Prompt Datasets
  18. Task Decomposition and Workflow Prompting
  19. Reasoning, Test-Time Compute, and Verifiers
  20. Context Windows, Position Effects, and Effective Recall
  21. Context Engineering: Selection, Ordering, and Compression
  22. Structured Outputs and Constrained Generation
  23. Function Calling and Tool Contracts
  24. Conversation State, Memory, and Privacy Boundaries
  25. Retrieval-Augmented Generation as an Evidence Pipeline
  26. RAG Corpus Design, Chunking, and Metadata
  27. Sparse, Dense, and Hybrid Retrieval
  28. Reranking and Context Assembly
  29. Grounding, Citations, and Claim-Level Support
  30. RAG Evaluation and Debugging
  31. Prompt, RAG, Tool, or Fine-Tune?
  32. Fine-Tuning Data Curation and Release Design
  33. Parameter-Efficient Adaptation: LoRA, QLoRA, and Adapters
  34. Synthetic Data and Model-Generated Supervision
  35. Multimodal Foundation Models
  36. Generative Media, Editing, and Provenance
  37. Evaluating Generative AI Systems
  38. Hallucination, Uncertainty, Verification, and Abstention
  39. Security, Privacy, and Abuse-Resistant Design
  40. Model Selection, Routing, and Economics
  41. Serving LLMs: Prefill, Decode, KV Cache, and Batching
  42. Monitoring, Feedback, and Change Management
  43. Red-Teaming, Incident Readiness, and Release Governance
  44. Generative AI Capstone: Design and Defend a Grounded Product