.

.

.

5

.

M

.

.

5

.

.

.

.

.

.

.

5

.

.

.

.

5

.

I gave this course 4 stars because, while I believe prompt engineering is gradually becoming less central as an explicit discipline, it still provides valuable insights. As generative AI models become increasingly capable of understanding natural language intuitively, and as they are augmented with retrieval mechanisms and agentic capabilities—as the course itself acknowledges—the need for highly structured prompt engineering is likely to diminish over time. That said, the course helped me better reflect on how these models process instructions and how important it is to communicate clearly and intentionally when interacting with them. Prompt engineering remains a valuable soft skill, because at its core it is about effective communication. Just as with humans, the quality of the input strongly influences the quality of the output. I also realized that many of the techniques and best practices presented in the course were things I had already been applying intuitively through daily use of generative AI, especially in coding and practical workflows. For experienced users, the course may not significantly change their approach or provide groundbreaking practical improvements. However, from a theoretical and conceptual perspective, it is still useful. It provides a structured framework for understanding why certain prompting techniques work and helps formalize practices that many experienced users develop organically over time. Overall, while it may not be transformative for advanced users, it remains a valuable course for strengthening conceptual understanding and improving intentional interaction with generative AI systems.

I enjoyed this course especially due to its lab environment which allows students to really get their hands dirty on prompting techniques. There are two reasons why I don't give the course 5 stars: 1) In the more complex prompting exercises using several experts (Bob puts a ball into a cup...), the provided sample prompts utterly failed in the requested choice of LLM. In some of these, the course description would say something like "with this approach, the LLM success rate will almost always be 100%". Not the case. But rewriting prompts and using a different LLM versions ultimately lead to some success. The Image Generation Module never worked once. When trying to create an image, the module would always trigger an error message of the type "This image expired." I created a ticket for this but in the end practiced on our company-owned Gemini version instead. Nevertheless, a good course but might soon need some updates since technology is moving fast!