Google Cloud · Generative AI Leader
Model limitations and how to address them
Generative AI Leader · Lesson 9 of 14 · 1:17:35
YouTube loads only after you press play. This chapter opens at its timestamp in the full course. Watch on YouTube ↗
In this lesson
Prepare for the Google Cloud Generative AI Leader exam. In this lesson: the limitations of foundation models (data dependency, knowledge cutoff, bias, fairness, hallucinations, edge cases), Google Cloud's recommended practices (grounding, RAG, prompt engineering, fine-tuning, humans in the loop), pre- and post-generation reviews, and continuous monitoring: KPIs, Model Registry, Model Monitoring for drift, Feature Store, model upgrades, and security patches. Includes three original practice questions.
Preview the topics
- Cymbal Retail’s assistant confidently tells a shopper about a promotion that never existed. Which limitation is this?
- Cymbal Retail’s assistant knows nothing about the product line launched last month, after the model’s training ended. What is the best fix?
- Before Cymbal Retail publishes product safety descriptions written by gen AI, experts must validate each one. Which practice is this?