Artificial Intelligence

Foundations Of AI

The AI/ML/DL/NLP Landscape

Chapter 3 introduced the symbolic/sub-symbolic distinction. This chapter makes the full landscape explicit — because "AI," "Machine Learning," "Deep Learning,"

JrCodex·7 min read

Jr Codex AI Notes

Level: Beginner–Intermediate Prerequisites: Chapter 4 Time to complete: ~20 minutes


Table of Contents

  1. Why This Chapter Exists
  2. The Nested Relationship
  3. Artificial Intelligence — the Umbrella
  4. Machine Learning — Learning from Data
  5. Deep Learning — Learning with Neural Networks
  6. NLP & LLMs — Language-Specific AI
  7. Data Science — a Related but Distinct Discipline
  8. How This Curriculum Is Organized
  9. Summary & Next Steps

1. Why This Chapter Exists

Chapter 3 introduced the symbolic/sub-symbolic distinction. This chapter makes the full landscape explicit — because "AI," "Machine Learning," "Deep Learning," "NLP," and "Data Science" get used interchangeably in casual conversation, but they refer to distinct (if overlapping and related) fields, each covered by its own dedicated set of notes in this curriculum.


2. The Nested Relationship

The Standard Nesting
─────────────────────────────────────────
  ┌───────────────────────────────────────────────┐
  │  Artificial Intelligence (this curriculum's      │
  │  primary subject: search, logic, planning,         │
  │  probabilistic reasoning, agents)                    │
  │                                                        │
  │   ┌─────────────────────────────────────────┐           │
  │   │  Machine Learning (learning from data)     │           │
  │   │  — dedicated Machine Learning notes           │           │
  │   │                                                 │           │
  │   │   ┌─────────────────────────────────┐             │           │
  │   │   │  Deep Learning (neural networks)   │             │           │
  │   │   │  — dedicated Deep Learning notes      │             │           │
  │   │   │                                         │             │           │
  │   │   │   ┌─────────────────────────┐             │             │           │
  │   │   │   │  NLP / LLMs                │             │             │           │
  │   │   │   │  (language-specific)          │             │             │           │
  │   │   │   │  — dedicated NLP/LLM notes       │             │             │           │
  │   │   │   └─────────────────────────┘             │             │           │
  │   │   └─────────────────────────────────┘             │           │
  │   └─────────────────────────────────────────┘           │
  └───────────────────────────────────────────────┘
─────────────────────────────────────────

Every inner circle is a subset of the outer one — every deep learning model is a machine learning model; every machine learning system is (by the broad definition from Chapter 1) an AI system. But not every AI system is machine learning — this is the single most important, most commonly confused point (also raised in Chapter 1).


3. Artificial Intelligence — the Umbrella

As established in Chapters 1-4: AI is the broadest field, covering any system that perceives and acts to achieve goals — including techniques that involve no learning from data whatsoever.

Classical (Symbolic) AI — This Curriculum's Focus
─────────────────────────────────────────
  Search algorithms (Module 2)        — finding a route, solving a puzzle
  Knowledge representation & logic       — encoding facts and rules (Module 3)
  Planning                                  — sequencing actions to reach a goal (Module 3)
  Probabilistic reasoning                    — reasoning under uncertainty (Module 4)
  Multi-agent systems                          — multiple agents interacting (Module 5)

  NONE of these techniques require training a model on data —
  they're built from explicit algorithms and encoded knowledge.
─────────────────────────────────────────

4. Machine Learning — Learning from Data

Machine Learning (ML) is the subfield of AI focused on systems that improve their performance by learning patterns from data, rather than following explicit, hand-coded rules.

# Classical AI (symbolic, hand-coded) — a spam filter using explicit rules
def is_spam_classical(email_text):
    spam_words = ["win money", "click here now"]
    return any(word in email_text.lower() for word in spam_words)
 
# Machine Learning (learned from data) — a spam filter TRAINED on thousands
# of labeled emails, discovering patterns a human never explicitly wrote
# (full depth in the Machine Learning notes)
# model.fit(training_emails, training_labels)
# model.predict(new_email)

The Machine Learning Notes cover this in full depth: data preprocessing, model evaluation, linear/logistic regression, decision trees, ensemble methods, clustering, and more — 12 chapters dedicated entirely to this subfield.


5. Deep Learning — Learning with Neural Networks

Deep Learning (DL) is the subfield of machine learning specifically using neural networks with many layers ("deep" networks) — the technique behind the 2012+ AI boom referenced in Chapter 2's history.

Why Deep Learning Is Its Own Dedicated Subfield
─────────────────────────────────────────
  Traditional ML often requires a human to manually engineer
  useful FEATURES from raw data (Data Science notes, Module 2,
  Ch.5) before a model can learn from them.

  Deep learning networks can learn USEFUL FEATURES directly
  from raw data (pixels, raw text, audio waveforms) themselves
  — this capability is what enabled the breakthroughs in image
  recognition, speech, and language covered in the dedicated
  Deep Learning notes.
─────────────────────────────────────────

The Deep Learning Notes cover neural networks from scratch, training deep networks, CNNs, RNNs/LSTMs, transformers/attention, transfer learning, and generative models — 10 chapters.


6. NLP & LLMs — Language-Specific AI

Natural Language Processing (NLP) is the subfield (spanning both classical and deep-learning techniques) focused specifically on understanding and generating human language. Large Language Models (LLMs) are the current, dominant deep-learning-based approach to NLP.

NLP's Own Internal History (Echoes Ch.2's Boom-Bust Pattern)
─────────────────────────────────────────
  Classical NLP:    rule-based grammars, bag-of-words, TF-IDF
                     (some overlap with THIS curriculum's symbolic
                      knowledge representation, Module 3)
  Statistical NLP:    probabilistic language models (connects to
                        Module 4's probabilistic reasoning)
  Deep Learning NLP:    word embeddings, RNNs, and eventually the
                          Transformer architecture (2017)
  LLM era:                GPT, Claude, and similar models — trained
                            on massive text corpora, capable of
                            fluent generation and broad task-following
─────────────────────────────────────────

The NLP & LLM Notes cover this full arc: classical NLP, word embeddings, BERT/GPT architectures, fine-tuning, prompt engineering, RAG, LLM agents, and evaluation — 12 chapters.


Data Science overlaps heavily with AI/ML but isn't a subset of it in the same nested way — it's better understood as a sibling discipline focused on extracting insight and answering questions from data, using statistics, visualization, and experimentation, with machine learning as just one of its tools.

AI/ML                                Data Science
─────────────────────────────         ─────────────────────────────
  Primary goal: build systems that      Primary goal: answer QUESTIONS
  PREDICT or ACT (often for                and inform DECISIONS from data
  production/deployment)                     (often for a specific business
                                                or research question)

  A trained model deployed to                A/B test showing a marketing
  score millions of users                      campaign increased sales 4%
─────────────────────────────         ─────────────────────────────

The Data Science Notes cover statistics/probability foundations, data wrangling, EDA, visualization, and experimentation (A/B testing) — 7 modules, and explicitly bridges into the Machine Learning notes once its own foundational work is done.


8. How This Curriculum Is Organized

This Curriculum's Five Notes Folders
─────────────────────────────────────────
  Python              → the programming foundation everything else assumes
  Data Science          → statistics, wrangling, EDA, visualization, experimentation
  Artificial Intelligence  → (THIS folder) classical/symbolic AI: search, logic,
                              planning, probabilistic reasoning, agents, RL basics
  Machine Learning           → learning from data: algorithms, evaluation, production
  Deep Learning                → neural networks specifically
  NLP & LLMs                     → language-specific AI, classical through modern LLMs
─────────────────────────────────────────

A practical reading order, if starting from zero: Python → Data Science → this Artificial Intelligence curriculum → Machine Learning → Deep Learning → NLP/LLMs. Each hands off to the next, and this AI curriculum's Module 7 closing chapter maps out the specific connections in detail once you've completed it.


9. Summary & Next Steps

Key Takeaways

  • AI is the broadest umbrella term; Machine Learning is a subset of AI (learning from data); Deep Learning is a subset of ML (using neural networks); NLP/LLMs are a subset of DL/ML applied specifically to language.
  • Not every AI system is machine learning — classical, symbolic techniques (search, logic, planning, probabilistic reasoning) build genuine AI without any data-driven learning at all, and are this curriculum's primary focus.
  • Data Science is a related but distinct sibling discipline — focused on answering questions from data using statistics and experimentation, with ML as one tool among several, not its defining purpose.
  • This curriculum splits across five dedicated notes folders (Python, Data Science, AI, Machine Learning, Deep Learning, NLP/LLMs), each building on the last.

Concept Check

  1. Why is it inaccurate to say "AI is just machine learning"?
  2. What capability does deep learning add on top of traditional machine learning?
  3. How does Data Science's primary goal differ from AI/ML's primary goal?

Module 1 Complete — Next Module

You now understand what AI is, where it came from, how it's classified, the agent framework every technique in this curriculum builds on, and exactly how this curriculum's classical AI focus relates to the ML/DL/NLP notes. Module 2 begins the technical core of classical AI: solving problems through search.

Next Chapter

Module 2: Problem Solving by Search


Jr Codex — 1-on-1 Personalized Coaching | Back to Module Index