Sunday, August 10, 2025

๐ŸŽ“ I just finished: Trustworthy Generative AI — Module 1 (Part 1)

I just finished the first part of Module 1 in Trustworthy Generative AI from Vanderbilt University (course #3 in Jules White’s Prompt Engineering sequence). This course rounds out my learning path after Prompt Engineering for ChatGPT and ChatGPT Advanced Data Analysis. It also contributes toward the specialization certificate that bundles all three courses. (Coursera)

Result: 100% on both graded assignments.

What this part covered

Module 1’s opening unit, “The Right Types of Problems to Solve with Generative AI,” reinforced good problem selection and risk-aware prompting. Highlights from my notes (mirroring the lesson list you see in the screenshot):

  • GenAI is not a source of facts — verify.

  • Prefer problems where correctness is easy to check; avoid hard-to-check answers.

  • Look for cases where partial answers still add value.

  • Think about risk and human oversight.

  • Does the use benefit you, the human?

  • Intro to the ACHIEVE framing used across Vanderbilt’s materials.

  • Two short, graded exercises (both completed with full marks).

Reflection

Nothing radically new for me, but it nudged my perspective on a few everyday uses of ChatGPT—especially framing tasks so that validation is easy and risk is low. It’s a solid start to a course that focuses on trust, verification, and appropriate human involvement, consistent with the course’s stated aims. (Coursera, Class Central)

No comments:

Post a Comment

๐Ÿง  Transfer of Consciousness: Moving Long-Running AI Projects Between Chats

When you work with conversational AI over time, you quickly discover an odd limitation: Your project stays in the old chat. The capabiliti...