Descriptions:
Andrew Ng, co-founder of DeepLearning.AI and one of the most recognized figures in AI education, presents a comprehensive 2026 prompting course covering how effective AI usage has evolved since ChatGPT launched in 2022. The course is structured around a recurring contrast between AI novices and power users, with concrete before-and-after examples drawn from tools including ChatGPT, Anthropic’s Claude, and Google Gemini.
Core techniques include: providing sufficient context by treating AI like a capable but uninformed new hire (uploading project trackers, voice memos, and documentation rather than writing short prompts), asking neutral questions to counteract the sycophancy problem where AI reflects back what it thinks you want to hear, and using explicit rubrics to force objective evaluation. Ng also covers context window management — deliberately curating what information is in the model’s context rather than flooding it — and explains when to let the model think at length versus when quick answers suffice.
The course extends into agentic desktop applications, where tools like Claude’s computer-use features, Microsoft Copilot, and Google’s equivalent can autonomously read, rename, and reorganize files on your computer with iterative human review. Ng walks through a file organization workflow end-to-end to illustrate the propose-then-confirm pattern that makes these tools practical. For anyone who learned to use AI in 2022 and hasn’t revisited their approach, this course offers a structured update covering the full capability surface of today’s models.
📺 Source: DeepLearningAI · Published May 18, 2026
🏷️ Format: Course Lesson
