Learning path · Level 1 · Tutorial 1

Distinguishing AI, Machine Learning, and Generative AI

Use everyday examples to distinguish automation, AI, machine learning, deep learning, and generative AI without treating them as interchangeable terms.

  • Level: Beginner
  • Duration: 12 min
  • 7 steps
Start tutorial Back to the path
Distinguishing AI, Machine Learning, and Generative AI

Step 1 of 7

Start with Fixed-Rule Automation

See how a directly configured schedule can produce repeated automatic behavior.

A rule can trigger an automatic action

Automation follows a setup

Automation carries out a process with less repeated human action. A person sets it up once; then the system performs the defined action when the relevant time or condition occurs.

In fixed-rule automation, the behavior comes from explicitly configured instructions. The system does not need to decide a new rule each time.

From configured rule to action

A visual flow from setting a clock to an automation process that sends a phone reminder.

A configured schedule leads the system to perform an action automatically.

The behavior is configured directly

Example: Reminder Timer

Consider Reminder Timer. A person explicitly configures this instruction:

“At 9:00 AM every day, send a reminder.”

After that setup, Reminder Timer checks the time and sends the reminder each day. The person does not have to press a button every morning.

Example

Trace the process

Configured instruction

Schedule: 9:00 AM every day
Action: send a reminder

The repeated behavior is specified directly by the person who configured the schedule.

Tip

Notice the two parts

A fixed rule can have a condition and an action. Here, reaching the configured time is the condition; sending the reminder is the action.

Practice: trace the configured rule

Put the rule-driven process in order

Put the stages of Reminder Timer in the order that explains why it sends a reminder automatically.

  1. Reminder Timer sends the reminder automatically.
  2. Reminder Timer checks the current time.
  3. The current time reaches the configured schedule.
  4. A person configures a daily 9:00 AM reminder rule.

Step 2 of 7

AI Is Broader Than Learning

Use mechanism-based evidence to distinguish fixed-rule automation from rule-based AI, without assuming that automatic systems are AI or that every AI system learns from data.

Look at the mechanism

AI is broader than automatic operation

Artificial intelligence (AI) is a broad field concerned with systems that perform tasks such as reasoning, recognizing patterns, and producing content.

AI does not describe one single way of building a system. Some AI systems use expert-written rules. Other AI systems learn patterns from data. A system operating automatically is not, by itself, evidence that it uses AI.

To label a system carefully, look for the mechanism described—not merely for an action that happens without repeated human input.

Two automatic systems, different evidence

A comparison between a direct scheduled reminder workflow and a plant-advice system that combines observations with expert rules.

The Reminder Timer follows a configured schedule directly. The Plant Adviser combines observations with expert-written rules to infer possible causes.

Compare the two systems

Rule-based AI does not have to learn

The Plant Adviser is a rule-based AI expert system. A person reports observations about a houseplant, such as yellow leaves or dry soil. The system combines expert-written rules to infer possible causes.

Its rules were supplied by experts; in this description, it does not learn those rules from example cases. It can still be AI because reasoning with expert rules is one approach within AI.

This differs from the Reminder Timer, whose person-configured schedule directly specifies when to send a reminder.

Evidence from the descriptions

Read the descriptions as evidence for a label. A label can be supported, or it can be not established when the description does not provide enough evidence.

What each description does—and does not—support
SystemDescribed mechanismAI label
Reminder TimerConfigured schedule sends a reminder at 9:00 AMNot established
Plant AdviserExpert-written rules combine observations to infer possible causesSupported

Challenge the shortcut

Check the evidence

True or false: If a system operates automatically, its description establishes that it uses artificial intelligence (AI).

Apply the comparison

Explain both distinctions

Reject both claims using the mechanisms described for the Reminder Timer and the Plant Adviser. For each claim, explain what the description does or does not establish.

  1. “The Reminder Timer acts automatically, so it must be AI.”
  2. “The Plant Adviser is AI, so it must have learned from data.”

Write at least 120 characters (0/120).

Step 3 of 7

Machine Learning Finds Patterns in Data

Distinguish explicit rules from patterns learned from data, and place machine learning within artificial intelligence.

From Examples to Useful Patterns

Machine Learning: Patterns from Data

Machine learning (ML) is an approach within artificial intelligence (AI) in which a system learns useful patterns from data.

Consider the fictional Spam Filter. It learns patterns from example emails to flag spam. A person does not need to write every possible rule for recognizing spam. The important distinction is how its task behavior was developed: patterns were learned from examples.

A comparison diagram: example emails help one system find patterns, while explicit rule cards direct another system.

The difference is how task behavior is developed: learned patterns versus rules written directly by people.

Compare the Evidence

Automatic Does Not Mean Data-Learned

Both kinds of systems can operate automatically after they are set up. Automatic operation does not tell you whether a person wrote the rules or whether the system learned patterns from data.

Ask this question instead: Was the behavior specified directly with rules, or developed by learning patterns from examples?

Compare the source of task behavior, not whether the system runs without repeated human action.

Two ways task behavior can be developed
SystemHow its behavior is developedCan it operate automatically?
Reminder TimerA person configures a schedule and reminder actionYes
Plant AdviserExperts write rules it combines to infer possible causesYes
Spam FilterIt learns useful patterns from example emailsYes

Tip

Key Distinction

A data-learned approach still runs as software and follows instructions when it operates. “Learned from data” describes how useful task behavior was developed; it does not mean the system acts without any software instructions.

Place ML Within AI

ML Is Part of AI

Machine learning is inside AI: it is one approach used within the broader AI field. But AI is broader than machine learning. For example, the Plant Adviser is rule-based AI; its expert-written rules do not learn from data.

So, an AI description may establish machine learning, or it may establish a different AI approach. Look for evidence of learned patterns before applying the machine learning label.

A large AI region contains a smaller ML region. A learned email-filter example is inside ML; a rule-based plant adviser is inside AI but outside ML.

Machine learning belongs within AI, while some AI uses expert-written rules rather than machine learning.

Apply the Distinctions

Check the Development Method

Match each description to the conclusion supported by its evidence.

Match each description with its supported conclusion.

Tap an item and then its matching pair.

Step 4 of 7

Place Deep Learning Inside Machine Learning

See the containment relationship

A nested hierarchy

Machine learning (ML) is part of artificial intelligence (AI) and learns patterns from data. Deep learning (DL) is a narrower approach within ML. It uses artificial neural networks with many processing layers.

Think in terms of containment: a deep learning system is automatically an ML system and an AI system. The reverse is not guaranteed.

Three nested regions show Deep Learning inside Machine Learning inside Artificial Intelligence, with a handwritten-digit reader in the deepest region.

Deep learning is contained within machine learning, which is contained within artificial intelligence.

Use the description, not an assumption

Example

Example: Digit Reader

Digit Reader recognizes handwritten digits. Its description explicitly says that it uses an artificial neural network with many processing layers. That evidence makes the deep learning label Supported.

Since deep learning is within machine learning and AI, Digit Reader is also an ML system and an AI system.

Read only what the evidence supports

Be careful with the direction of the inference. If a description says only that a system uses ML, deep learning is Not established. It may use deep learning, but the description has not supplied the needed evidence.

A broad label does not automatically prove a narrower one.

Practice the hierarchy

Check the direction

Which conclusion is justified when a system is explicitly described as using deep learning?

Final check

Complete the containment statement

Complete the statement: Deep learning is a subset of _.

Identify the necessary broader labels

A system explicitly described as using deep learning must also have which two broader labels?

Step 5 of 7

Generative AI Describes a Capability

Recognize generative AI as a content-producing capability that can overlap with machine learning and deep learning.

Recognize the capability

A capability, not a separate layer

Generative AI describes a capability: producing new content in response to an input. That content can be text, images, audio, or video.

This is different from a label for how a system was built. Machine learning (ML) and deep learning (DL) are approaches used to build systems. Generative AI tells you what a system can do: create content.

What content generation looks like

A request leads to four types of newly created content: a story, an image, audio, and video.

Generative AI can produce different forms of content from an input.

Attach labels from the description

Same approach, different capability

Consider two systems whose descriptions give different evidence:

  • Story Writer uses deep learning to draft a short fictional story from a request. It produces new text, so it has the generative AI capability. Because its description explicitly says it uses deep learning, it is also DL, ML, and AI.
  • Digit Reader uses deep learning to recognize handwritten digits. Its description establishes DL, ML, and AI—but it does not generate new content.

So, deep learning does not automatically mean generative AI.

Compare the evidence

One deep-learning system creates a story from a request, while another recognizes a handwritten digit.

Both examples use deep learning; only Story Writer is described as producing content.

Example

Story Writer’s compatible labels

Labels can overlap.

For Story Writer:

  • AI: Supported
  • ML: Supported
  • DL: Supported
  • generative AI: Supported

Generative AI is best pictured as a capability badge attached to an applicable system, not as a new region outside or above AI, ML, and DL.

Check the capability

Practice: spot content production

Which description provides evidence of generative AI?

Explain the overlap

Apply the distinction

Why can Story Writer correctly be labeled both deep learning and generative AI?

Step 6 of 7

Choose Labels from the Evidence

Use stated clues to assign compatible labels while leaving unsupported details unresolved.

Read descriptions as evidence

Follow the evidence, not the label

When you label a system, use only what its description actually establishes. Labels can stack rather than compete.

A stated learning-from-data clue supports machine learning (ML), and ML is inside artificial intelligence (AI), so AI is supported too. A stated artificial neural network with many processing layers supports deep learning (DL), which also supports ML and AI.

Likewise, a clue that a system produces new content supports both generative AI and AI. If a description lacks a narrower clue, mark that label Not established. That does not mean the label is false; it means the description does not justify it. A familiar product name or an “AI-powered” claim does not establish a particular learning method.

Evidence can support different labels

An evidence diagram with icons for schedules, rules, example messages, many network layers, and generated writing leading to nested category shapes and a separate capability badge.

Look for the implementation or capability clue: rules, learned patterns, many layers, and newly produced content each support different labels.

Keep supported and unknown separate

Evidence table

Use a compact evidence table to keep supported labels separate from details that are not established. A label may be supported because it is directly stated or because it follows from the AI–ML–DL hierarchy.

What each clue supports

“Not established” means the description does not provide enough evidence for that label.
Description clueSupportedNot established
Expert-written rules infer possible plant causesrule-based AI; AIML; DL; generative AI
Learns patterns from example emailsML; AIDL; rule-based AI; generative AI
Artificial neural network with many layers recognizes digitsDL; ML; AIrule-based AI; generative AI
Drafts a new fictional story from a request; method not statedgenerative AI; AIML; DL; rule-based AI
Configured schedule sends a reminder at 9:00 AMfixed-rule automationAI; rule-based AI; ML; DL; generative AI

Check the clues

Practice: match clues to labels

Match each system description with the evidence statement it justifies.

Tap an item and then its matching pair.

Make an evidence-based judgment

Apply all justified labels

A Message Sorter learns patterns from thousands of previously categorized messages. It automatically routes each new message to a folder. Its description does not mention an artificial neural network with many processing layers, explicitly configured rules, or producing new content.

In 3–5 sentences, identify every supported label and every relevant label that is Not established. Cite the clue for each conclusion.

Write at least 120 characters (0/120).

Step 7 of 7

Apply the Distinctions

Use evidence from unfamiliar scenarios to apply the distinctions among automation, AI, machine learning, deep learning, and generative AI.

Apply an evidence checklist

Use the clues in the description

Use the description as evidence—not a product name or the fact that something runs automatically.

  • An explicitly configured condition or schedule supports fixed-rule automation.
  • Expert-written rules that combine observations to infer an answer support rule-based AI.
  • Learned patterns from examples support machine learning (ML), which is within artificial intelligence (AI).
  • A many-layer artificial neural network supports deep learning (DL), which is within ML.
  • Producing new content from a request supports generative AI.

A label can be Supported, or Not established when the description does not provide enough evidence. Labels can overlap; they are not a forced single choice.

Map the evidence to labels

A diagram separates fixed-rule automation from nested AI, ML, and DL regions, with generative AI shown as a capability badge.

Read labels from the evidence supplied: DL is within ML, which is within AI; generative AI is a capability that can overlap with them.

Summary

Final evidence checklist

  • Automatic operation by itself does not establish AI.
  • ML is within AI; DL is within ML.
  • Generative AI describes a content-producing capability, not a separate required level in the hierarchy.
  • Use “Not established” when a narrower label lacks evidence.

Check a configured process

Scenario 1: An explicit rule

A warehouse manager configures Shelf Signal: “When the count of item A falls below 12, send a reorder alert.” Shelf Signal sends the alert without the manager repeating the action.

Which conclusion is best supported?

Check containment and capability

Scenario 2: Overlapping labels

Invite Studio learns visual patterns from many example invitations. Its artificial neural network has many processing layers. When a user describes an event, it creates a new invitation image.

Which set of labels is Supported?

Explain, do not just label

Final application: justify the labels

Write a short evidence-based analysis of all three scenarios below. For each one, name the labels that are Supported and identify any narrower labels that are Not established. Cite the clue that led you there.

  1. Gatekeeper checks each visitor badge against access rules written by security staff. It combines those rules to infer whether a visitor may enter. Its rules are not updated from examples.
  1. Receipt Sorter learned patterns from thousands of labeled receipt images and now places new receipts into expense folders. Its architecture is not described.
  1. Caption Composer uses an artificial neural network with many processing layers. From a request for a product caption, it writes a new caption rather than selecting a fixed prewritten message.

Finish with one sentence explaining why the terms are related but not interchangeable.

Write at least 350 characters (0/350).

Tutorial completed

Congratulations! You completed: Distinguishing AI, Machine Learning, and Generative AI

You can now distinguish fixed-rule automation, rule-based AI, machine learning, deep learning, and generative AI by the evidence in a description.

Download the App now to have access to + 5000 free courses, exercises, certificates and lots of content without paying anything!

  • 100% free online courses from start to finish

    Thousands of online courses in video, ebooks and audiobooks.

  • More than 60 thousand free exercises

    To test your knowledge during online courses

  • Valid free Digital Certificate with QR Code

    Generated directly from your cell phone's photo gallery and sent to your email

Cursa app on the ebook screen, the video course screen and the course exercises screen, plus the course completion certificate