AIF-C01 - Generative AI fundamentals (24% of the exam) - Section 2.3
Understand the capabilities and limitations of generative AI for business problems.
Weigh strengths (drafting, summarisation, code assistance) against limitations (hallucination, lack of grounding, cost, and inconsistency). Judge whether generative AI is the right tool for a given task.
HallucinationGroundingCost and latency
Practice question for this objective
lock_openFree sampleGenerative AI fundamentalsmedium
An insurance firm wants to use a generative AI model to answer customer queries about policy terms. The model frequently gives responses that are fluent but contradict the actual policy documents. Which mitigation technique directly addresses this limitation by anchoring model outputs to verified source material?
AFine-tuning the model on a large corpus of general insurance industry articles
BIncreasing the model's temperature parameter to encourage more creative responses
CApplying retrieval-augmented generation so the model grounds its answers in retrieved policy documentscheck_circle Correct
DSwitching from a transformer-based model to a rule-based expert system for all responses
Understand how grounding techniques such as retrieval-augmented generation reduce hallucination by anchoring generative outputs to verified source documents. Grounding connects model inference to an authoritative external knowledge base retrieved at query time. Instead of relying solely on patterns encoded during training, the model receives the relevant document passages as context and is instructed to base its answer on them. This substantially reduces fabrication of policy-specific details while preserving natural language fluency.
Why A is wrong: Fine-tuning on broad industry articles updates model weights but does not guarantee outputs reflect the firm's specific policy documents; hallucination on specific policy details can persist.
Why B is wrong: Higher temperature increases output diversity and randomness, which would worsen factual accuracy, not improve it.
Why C is correct: Grounding via retrieval-augmented generation supplies the model with the authoritative source text at inference time, constraining outputs to verifiable content and directly countering hallucination on domain-specific facts.
Why D is wrong: A rule-based expert system is a valid approach for narrow decision trees but abandons the natural language capability the firm needs; grounding the existing model is a more targeted and practical mitigation.
Answers that look right on this material and are not. Each one is a distractor from a different question in the AIF-C01 bank for this domain.
Lack of multimodal capability, because the model cannot process scanned PDF images of case documents
errorWhy it is wrong: Multimodal capability is a separate architectural concern; the partner's worry is about accuracy of text summaries, not the ability to process different input formats.
The size of the training dataset used during model pre-training
errorWhy it is wrong: Training dataset size affects model quality but is entirely irrelevant to inference latency; it has no bearing on how quickly the model responds to a query after training is complete.
Generative models require internet access to retrieve live tracking data, which introduces unacceptable latency
errorWhy it is wrong: Foundation models do not inherently require internet access; inference operates on provided context. This is not a characteristic that distinguishes generative AI from time-series models for forecasting.