Examworthyexamworthy.comAWS Certified AI Practitioner cheat sheet
Amazon Web Services
Free to share. Examworthy is not affiliated with or endorsed by Amazon Web Services; AIF-C01 and related marks belong to their respective owners.
At a glance
Format: Multiple choice and multiple response
Domain weight map
Heaviest first - spend your time hereHow this exam thinks
AIF-C01 rewards judgement about where AI, ML, and generative AI each fit, not the ability to build or train a model.
Spot the trap
Tempting wrong answers, and why they failTempting but wrong
A model's training data cutoff date is what limits whether it can process a large document supplied at inference time.
Why it fails
Confuses baked-in knowledge with input capacity. Training cutoff affects facts stored in the weights, not the ability to process tokens supplied at inference. Providing the document as context bypasses the cutoff entirely; the limiting factor for input is the context window.
Applications of foundation models
Tempting but wrong
To make an LLM's repeated answers more consistent, raise the temperature toward 1.0 so it considers a wider spread of next tokens.
Why it fails
Wrong direction. Higher temperature flattens the token distribution and increases variation, moving away from on-message consistency. To reduce variation you lower temperature, which sharpens the distribution toward the likeliest tokens.
Generative AI fundamentals
Tempting but wrong
A rule-based expert system that encodes domain knowledge as hand-crafted if-then rules can improve its recommendations autonomously as new data arrives.
Why it fails
Category error. Rule-based systems require explicit programming of every new rule and do not learn autonomously from data. The ability to improve from experience without re-programming is the defining property of machine learning, not of rule-based expert systems.
Fundamentals of AI and ML
Tempting but wrong
Raising the temperature makes a model classify edge-case inputs more accurately because it explores more options.
Why it fails
Wrong lever. Higher temperature increases randomness, which worsens classification consistency rather than improving it. It also does not address the root cause of misclassification - the model's ambiguity about which label to assign.
Prompt engineering and evaluation
Tempting but wrong
Raising the model's overall accuracy on the full training dataset will fix bias against a disadvantaged subgroup.
Why it fails
Wrong target. Aggregate accuracy can rise while bias against a subgroup worsens, especially when that group is a small fraction of the data. Fixing disparity needs disaggregated per-subgroup evaluation and targeted mitigation, not a higher overall accuracy number.
Responsible and secure AI
Tempting but wrong
Setting temperature to zero (greedy decoding) reliably prevents a model from ever producing specific prohibited words.
Why it fails
Greedy decoding picks the most probable token at each step but cannot guarantee exclusion of specific words. A prohibited term may still appear if it is highly probable in context. Deterministic exclusion requires guardrails, not a temperature setting.
Applications of foundation models
Tempting but wrong
Top-p works by reweighting every token's probability by a scaling factor before sampling.
Why it fails
That describes temperature, not top-p. Reweighting the whole distribution by a factor is temperature scaling; top-p instead applies a cumulative-probability cutoff that keeps only the smallest set of top tokens summing to p.
Generative AI fundamentals
Tempting but wrong
Robotic process automation counts as machine learning because it automates repetitive human tasks.
Why it fails
Wrong. RPA follows a fixed, predefined script to mimic human steps and does not learn from data. Being automation does not make something ML; only systems that improve predictions from experience qualify as machine learning.
Fundamentals of AI and ML
Key terms
Exam-day rules
- Read the last line of the question first. It tells you what is actually being asked, so you can read the scenario looking for the answer rather than memorising detail.
- Choose the most appropriate option, not merely a correct one. Several options are often true; the exam wants the best fit for the stated requirement.
- Watch for absolutes such as always, never, and guarantees. In AI scenarios they are usually the wrong answer because models are probabilistic.
- Flag and move on. Do not lose time on one hard item when easier marks are waiting; the timer rewards covering every question first.
- When a source changes often, think retrieval and re-index before fine-tuning. This single distinction resolves a surprising number of Domain 3 questions.
Revision schedule
- Day 1Map the blueprint and set a date
- Week 1Lock the fundamentals (Domain 1)
- Weeks 1-2Go deep on generative AI and foundation models (Domains 2 and 3)
- Weeks 2-3Secure prompting and responsible AI (Domains 4 and 5)
- Week 4Practise on scenarios with worked explanations