Fundamentals of Generative AI
Generative AI Risks and Limits
Generative AI systems can produce incorrect, variable, costly, or unsafe outputs. AIF-C01 expects you to understand hallucination, grounding, non-determinism, latency, and token usage as design concerns.
Concepts
- Hallucination is confident but incorrect or unsupported output.
- Grounding ties model output to retrieved or supplied source material.
- Non-determinism means identical prompts may produce different outputs unless parameters are constrained.
- Latency and token usage affect user experience and cost.
Exam tips
- Grounding with retrieved or supplied context can reduce unsupported answers.
- Constrained inference settings can reduce variation but do not guarantee truth.
- Latency and token usage affect user experience and cost.