What's changed: In-scope coverage: added a data/infrastructure in-scope services block to Ch1 §2 (Athena/Glue/EMR/Lake Formation/Redshift/OpenSearch/QuickSight/Data Exchange/Lambda/ECS/EKS/EC2/RDS/Aurora/ElastiCache/Neptune/DocumentDB/DynamoDB/CloudFront)
1.1AI, Machine Learning, and Deep Learning
Understand how AI, machine learning (ML), and deep learning relate, and their basic concepts. The starting point for AIF-C01.
AI is the broadest concept; within it is machine learning (ML), and within that is deep learning. AIF-C01 (AWS Certified AI Practitioner) is an entry-level credential testing a map of "which technique exists for what purpose and which AWS service delivers it," more than deep math or coding. Solidifying this nesting and core ML terms is the starting point. Generative AI (later chapters) also builds on this deep-learning foundation.
1.1.1How they nest
- AI: an umbrella term for software performing human-like intelligent tasks.
- Machine learning (ML): a key AI approach that learns patterns from data to predict.
- Deep learning: an ML subfield using multi-layer neural networks to learn complex patterns.
1.1.2Core machine-learning terms
In ML, each data point is described by features (inputs) and a label (the answer). For house-price prediction, size and age are features, the actual price the label. Learning rules from data is training; using the trained model to answer on new data is inference. The key goal is generalization—being right on new, unseen data; the opposite failure, fitting training data too closely and missing on new data, is overfitting, the top concern in ML. So data is split into training and evaluation sets, with performance measured on the evaluation set.
Deep learning stacks neural networks (loosely modeling brain neurons) into many layers. Depth lets it progressively learn useful patterns even from complex data like image pixels or word sequences. The high accuracy of today’s image/speech recognition and generative AI rests largely on it. AIF-C01 tests no math—knowing "deep learning = multi-layer neural nets, strong on complex data, the basis of generative AI" suffices.
| Concept | Meaning | Example |
|---|---|---|
| AI | Software doing intelligent tasks (broadest) | Includes rule-based logic |
| Machine learning | Learn patterns from data to predict | Demand forecast, classification |
| Deep learning | Complex learning via multi-layer NN | Image recognition, LLMs |
| Generative AI | Create new content | Text/image generation |
Scenario: tell techniques apart by use. "Forecast next month’s sales from history" = ML (supervised regression); "recognize defects in many images" = deep learning (images); "auto-draft replies to inquiries" = generative AI. All are AI broadly, but the right method differs. AIF-C01 tests mapping "scenario → the right technique/AWS service."
Watch the mix-ups: (1) ML (predict/classify/cluster = pick an existing answer / estimate a number) vs generative AI (create new content)—different aims. (2) Deep learning is part of ML, not separate. (3) Don’t confuse features (inputs) vs labels (answers) or training (learn rules) vs inference (produce answers).
Q. Are AI and ML the same? No—AI is broader; ML is a key way to realize it (rule-based AI also exists). Q. What is overfitting? Fitting training data too closely so performance drops on new data—a failure of generalization. Q. Training vs inference? Training builds rules from data; inference returns predictions using the trained model.
Common: the nesting AI ⊃ ML ⊃ deep learning, core terms features/labels, training/inference, generalization/overfitting, and the ML (predict) vs generative AI (create) difference. Deep learning is part of ML and underpins generative AI (LLMs).
1.1.3Section summary
- Nesting AI ⊃ ML ⊃ deep learning; deep learning underpins generative AI
- Core terms: features/labels, training/inference, generalization/overfitting
- Distinguish by aim: ML = predict/classify, generative AI = create new content
Sign in to track progress — Log in.
Quick check
(just a quick review)Q1. Which correctly states the nesting of AI, ML, and deep learning?
Q2. Which field learns complex patterns with multi-layer neural networks?
Q3. Which best describes machine learning (ML)?