Until 2025, AI appeared in these exams as one or two general MCQs. In 2026 it became a full subject: the BSCS AP paper of 31.10.2026 had four 10-mark written questions on modern AI (NLP, generative AI/LLM, agentic AI with MCP, RAG), the AP paper of 17.01.2026 had two (A* search; supervised vs unsupervised vs reinforcement learning), and the 2026 screening MCQ paper (Set-3) asked about ensembles, hyperparameters, the dummy classifier, bias–variance and K-means. Every IT track now needs the same core, so it is collected here once instead of being spread over the other books.
Ch 2, 4 (AI in networks: anomaly detection, forecasting)
Ch 5, 6
How to write a 10-mark AI answer
Definition (1–2 lines) → a labelled diagram (pipeline, architecture or loop) → 4–6 key points or steps, each with one line of explanation → a comparison table where it fits (e.g. RAG vs fine-tuning) → one Bangladesh banking/government example → advantages and challenges → a one-line conclusion. Examiners reward structure and correct terms more than length.