Machine Learning vs. Artificial Intelligence: Key Differences

UPI Study › Courses › Introduction to Artificial Intelligence › Free lessons › Lesson 1.8 Free with a free account

Machine Learning vs. Artificial Intelligence: Key Differences

Introduction to Artificial Intelligence · Chapter 1: Foundations of AI and Core ConceptsSection 1.8 · lesson 8 of 6913 min · free with a free account

← Measuring Data Quality: Precision, Accuracy, and PitfallsNext lesson: Learnable Intelligence: Insights from Perkins' Framework →

Earn 3 college credits for Introduction to Artificial IntelligenceRecommended for credit by NCCRS · Computer Science 373 · self-paced · one payment, lifetime access
Enroll in Introduction to Artificial Intelligence · $250 →Download the syllabus (PDF)

In this lesson

How machine learning fits inside AI and when to use each approach.

  • Distinguish AI as a goal from machine learning as a method
  • Describe rule-based, search, and logic approaches that achieve AI without learning
  • Explain how deep learning differs from traditional machine learning
  • Choose between rules, search, machine learning, and deep learning for a given problem

Free study notes

Get this lesson's notes by email

The key points and 4 practice questions with answers, in your inbox. One email, free.

One email. Unsubscribe in one click.

Transcript

Everything said in the video, section by section (1,804 words). Read along, or search it.

Opening Question

Two Words Used as One

A phone ad promises AI. A job posting asks for machine learning. A headline uses both words for the same chatbot. People use AI and machine learning as if they meant the same thing. They're related, but they aren't identical. So what exactly is the difference? And when does it matter which one a system uses?

Analogy

The Goal Versus One Way There

Think about getting around a city. Getting people across town is the goal. A car is one good way to do it. But buses, trains and bicycles work too. In the same way, artificial intelligence is the goal, and machine learning is one very successful way of reaching it.

Definitions

Defined by What Versus by How

Artificial intelligence is the effort to build machines that do tasks we'd normally say need intelligence. Recognizing speech, planning a route, playing a strategy game, answering a question. AI is defined by what the machine achieves, not by how it achieves it. Machine learning is a method in which a computer improves at a task by studying examples, instead of being told every step by a programmer. So machine learning is defined by how the machine gets its ability: it learns it from data.

Nested Circles

AI Contains Machine Learning Contains Deep Learning

Picture the relationship as circles, one inside another. Artificial intelligence is the biggest circle. Machine learning sits inside it. And deep learning, which we'll meet later, sits inside machine learning. So every machine learning system is an AI system. But not every AI system uses machine learning. AI answers the question: what do we want the machine to do? Machine learning answers: how can it get that ability from data?

Rule-Based Systems

If-This-Then-That Knowledge

For decades, AI was built with no learning from data at all. People wrote the knowledge down, and many such systems are still in use, because they're reliable and easy to check. Let's meet the common families. The first is rules. A person writes if-this-then-that statements, and the program applies them to the situation. A tax tool that applies the filing rules to your income figures is a rule-based system. It's easy to read, check and correct. But it breaks down when there are too many rules, or the situation is unclear.

Search

Trying Actions to Reach a Goal

The second family is search. The program tries possible actions, one after another, to find a sequence that reaches a goal. A navigation app finding the shortest route between two places is doing search. Search can find the best answer when the problem is well defined. But the number of possibilities can explode.

Logic

Conclusions That Must Follow From Facts

The third family is logic. Facts are stored in a precise form, and the program draws conclusions that must follow from them. From stored facts and company policies, a logic system can conclude that a certain part cannot be shipped. Its conclusions can be explained step by step. But it needs facts that are complete and exact.

Clear Case

Perfect Tic-Tac-Toe With No Learning

Here's a clear case. Take a program that plays tic-tac-toe perfectly. The board is small, the rules are fixed, and the program can look at every possible game before it moves. It never learns anything from past games. Yet most people would agree it behaves intelligently. That's AI without machine learning.

Limits of Rules

Writing Rules for a Cat Photo

Rules work well when a person can describe the task precisely. But many tasks are very hard to describe. Try writing exact rules that tell a computer whether a photo shows a cat. You'd need rules for fur patterns, ear shapes, poses, lighting, and thousands of other variations. And you'd still miss many cats.

Learning From Examples

Labeled Photos Become a Model

Machine learning takes a different path. Instead of writing rules, we give the computer many examples with known answers: photos labeled cat, or not cat. A learning algorithm finds the patterns that separate the two, and stores what it found in a model. The model can then be applied to new photos it has never seen. So with rules, people supply the knowledge. With machine learning, the computer produces it from examples.

Worked Example

Rule-Based and Learning Spam Filters

A spam filter makes the contrast concrete. The rule-based version gets its knowledge from a person, who lists spam signs, like certain words, or too many links. The learning version studies thousands of emails already marked spam or not spam. When spammers change tactics, a person must notice and write new rules. The learning filter can be retrained on fresh examples. Each has its own risk. Spam that avoids every listed sign gets past the rules. The trained model may copy mistakes, or unfairness, from its examples. Notice that both filters do the same job. The goal hasn't changed. Only the way of getting the ability has.

Quick Check

Two Beach and Snow Photo Apps

Quick check. Two photo apps sort pictures into beach and snow. The first studied twenty thousand labeled photos. The second uses one rule a developer wrote: if more than half the pixels are white, call it snow. Which statement is right? Pause, and decide. Both are AI, and only the first one learned. Both do a task we'd call intelligent, so both count as AI. But only the first got its ability from data. The second runs a rule a person wrote, so it isn't machine learning. Option A is tempting, because the second app seems too simple. But AI is defined by what a system does, not by how it got there.

Side by Side

Written Knowledge Versus Extracted Knowledge

Set the two broad approaches side by side. Without learning, knowledge is written by people as rules, search steps or facts. No data is needed to build it, and you can usually trace why it decided. With learning, knowledge is extracted from example data, so the quantity and quality of data shape the results. And the learned patterns can be hard to interpret. Messy, fuzzy input defeats rules unless every case was foreseen, while a learned model often copes, if its examples covered similar cases. To adapt, people update the rules by hand, while a model is retrained. And the effort goes into writing knowledge, or into collecting, cleaning and labeling data.

Misconceptions

Three Myths About Machine Learning

Three misconceptions are worth clearing up. First: AI and machine learning mean the same thing. They don't. AI is the broader goal, and machine learning is one method. Second: machine learning is always better. Not so. For a task with clear, stable rules, a rule-based system can be cheaper, faster, and easier to trust. Third: learning means the machine understands. A learned model captures statistical patterns in its examples. That's powerful, but it isn't human understanding.

Deep Learning

Many Layers Building Complex Patterns

Now, deep learning. It's a branch of machine learning, not a separate field. It uses neural networks with many layers, which is where the word deep comes from. Each layer transforms its input a little, and passes the result on, so simple patterns combine into more complex ones. For a photo, early layers may pick out edges, middle layers shapes, and later layers whole objects.

Feature Design

Hand-Chosen Features Versus Raw Input

The most important practical difference is how the input is prepared. In traditional machine learning, people often decide in advance which measurable features matter, like the size or color of an object. Deep learning can work out useful features by itself, from raw input like pixels or sound waves. It usually needs much more data, and heavy computing, often on specialized hardware. It's strongest at images, sound and language. Traditional methods shine on table-like records.

Data and Performance

When Deep Learning Pulls Ahead

So is deep learning always the best choice? Here's a common pattern, as a sketch, not measured data. With little data, simpler methods often match deep learning, or beat it. As data grows, deep learning tends to pull ahead, if there's enough computing power to use it.

Choosing an Approach

Five Factors to Weigh

So how do you choose? Skilled practitioners start from the problem, and work toward the method. Five factors matter. Can the task be described exactly, with clear, stable rules? That favors rules. If it's easier to show examples than to explain, that favors learning. Is there example data with answers? Plenty, of good quality, favors learning. Little or none favors rules or search. Must every decision be explained, as in legal or safety settings? Does the situation change often? And what does a mistake cost? When exact guarantees are needed, rules win. When change is constant and occasional errors are tolerable, learning wins.

Decision Path

Questions That Lead to a Method

Those factors become a path of questions. First: can you write down the rules clearly? If yes, build a rule-based or logic system. If not: is the task finding a sequence of steps to a goal? If yes, use a search method. If not: do you have plenty of example data with answers? If not, gather more data, or go back to rules or search. If you do, ask one more question. Is the input raw images, sound or free text, with lots of data? Then deep learning. Otherwise, traditional machine learning.

Worked Examples

Benefits, Puzzles, and Card Fraud

Try the path on three problems. Deciding whether a person qualifies for a fixed government benefit, using written eligibility rules? The rules are exact, stable, and must be explained to the applicant. That's a rule-based system. Finding the quickest sequence of moves to solve a sliding-tile puzzle? A clear goal, well-defined moves, no examples needed. That's search. Flagging suspicious card payments, from millions of past transactions marked fraud or genuine? Subtle, shifting patterns, and plenty of labeled data. That's machine learning.

Quick Check

Reading Handwritten Addresses

Quick check. A postal service wants to read handwritten addresses on scanned envelopes. It has millions of past scans, each with the address already typed in by a clerk. Following the path, which approach fits best? Pause, and decide. Deep learning. Nobody can write clear rules for every handwriting style, and this isn't a search for steps. There's plenty of labeled data, and the input is raw images. That last question points to deep learning. Traditional machine learning is tempting, because the data has answers. But people would have to design features for handwriting, and raw images are where deep learning is strongest. Turning recorded speech into text lands in the same place: raw sound, endlessly varied, and huge numbers of recordings.

Combined Systems

Driver Assistance Uses Several Tools

Real systems often combine approaches. Take a driver-assistance system. It might use deep learning to recognize lane markings in camera images, search to plan a safe path, and fixed rules to guarantee it always brakes when an obstacle is very close. The skill lies in using each tool where it's strongest.

Key Takeaways

Goal, Method, and Choosing Wisely

To sum up. AI is the goal of intelligent machine behavior, and machine learning is one way to reach it. Rules, search and logic produce AI with no learning, and they're still valuable. Machine learning builds its ability from example data, and deep learning is the many-layered kind that shines on images, sound and language. Choose by asking whether rules can be stated, whether it's a search, and whether suitable data exists. And many systems combine all three.

Reflection

Rules or Learning for Hospital Triage

Here's something to think about. A hospital wants software that decides which patients need urgent attention. One team proposes hand-written rules. Another proposes a model trained on past patient records. What would you want to know before choosing? And could the two be combined?

About Introduction to Artificial Intelligence at UPI Study

Introduction to Artificial Intelligence (Computer Science 373) is a self-paced online course of 69 video lessons in 9 chapters, from intelligent agents and search to game strategy, neural networks, logic and reasoning, machine learning, language, robotics and AI ethics. It carries a NCCRS credit recommendation for 3 semester credits.

Every lesson transcript is free on pages like this one, and the videos are free with a free UPI Study account. Enrolling adds a practice quiz after each lesson, the graded work and final exam that earn the credit, lifetime access, and an official transcript on completion. UPI Study has helped over 130,000 students with college credits. ACE and NCCRS have over 2,100 cooperating colleges that will consider accepting credits; the final decision is with the universities.

Enroll now → Course details Download the syllabus All Computer Science courses Every course with PRO →

Compare: AI vs Current Trends in CS and IT

Why take it here

Why students take Introduction to Artificial Intelligence with UPI Study

How machines search, reason and learnIntroduction to Artificial Intelligence covers how AI systems are built: intelligent agents, search and heuristics, game strategy, logic, neural networks, machine learning, language and perception, and the ethics of AI. It is a common upper-level course in computer science degrees.
Credit at a fraction of the priceThe same 3 credits can cost $750 or more at a US college. Here it is one payment, self-paced, with lifetime access.
Finish your degree soonerTransfer credits reduce the courses you still owe, so you can graduate earlier or lighten a heavy semester.
Strengthen an applicationA completed college-level course on an official transcript shows admissions and scholarship committees you can do the work.
Study on your own scheduleNo start dates, no deadlines. Watch the lessons, practice, and take the final when you are ready.
130,000+students helped with college creditsACE and NCCRS have over 2,100 cooperating colleges that will consider accepting credits; the final decision is with the universities.

Where the credit goes

Universities that work with NCCRS

Introduction to Artificial Intelligence carries a NCCRS credit recommendation. These universities work with NCCRS; some are UPI Study's signed partners. ACE and NCCRS have over 2,100 cooperating colleges that will consider accepting credits; the final decision is with the universities.

Thomas Edison State UniversityThomas Edison State UniversitySigned partnerUniversity of the PeopleUniversity of the PeopleSigned partnerCharter Oak State CollegeCharter Oak State CollegeSigned partnerStrayer UniversityStrayer UniversitySigned partnerColumbia Southern UniversityColumbia Southern UniversitySigned partnerIndiana TechIndiana TechSigned partnerPenn StatePenn StateWorks with NCCRSSouthern New Hampshire UniversitySouthern New Hampshire UniversityWorks with NCCRSSUNYSUNYWorks with NCCRS +many more cooperating universitiesFind my college →

Find my college → Not listed? Ask Athena in the corner of this page, or book a 15-minute call.

FAQ

Questions students ask us

About Machine Learning vs. Artificial Intelligence: Key Differences

What is the difference between AI and machine learning?
Artificial intelligence is the broad goal of building machines that do tasks normally needing intelligence, while machine learning is one method in which a computer improves at a task by studying examples. AI is defined by what a machine achieves, and machine learning by how it gets that ability. Every machine learning system is an AI system, but not every AI system learns from data.
Can a system be AI without using machine learning?
Yes, rule-based systems, search, and logic can produce intelligent behavior without any learning from data. A program that plays tic-tac-toe perfectly by examining every possible game never learns from past games, yet most people would agree it behaves intelligently. A tax tool that applies filing rules and a navigation app that searches for the shortest route are further examples of AI without machine learning.
How is deep learning different from traditional machine learning?
Deep learning is a branch of machine learning that uses neural networks with many layers, and it can work out useful features by itself from raw input like pixels or sound waves. Traditional machine learning often relies on people deciding in advance which features matter. Deep learning usually needs much more data and computing power and is strongest on images, sound, and language, while traditional methods shine on table-like records.
When should you use rules instead of machine learning?
Rules fit best when a task can be described exactly with clear, stable rules, when every decision must be explained, or when little labeled data exists. Deciding eligibility for a fixed government benefit from written rules is a good example. Machine learning suits tasks that are easier to show by example than to explain, such as flagging card fraud from millions of past transactions marked fraud or genuine.

About the Introduction to Artificial Intelligence course

How many college credits is Introduction to Artificial Intelligence?
Introduction to Artificial Intelligence (Computer Science 373) carries a NCCRS credit recommendation for 3 semester credits. ACE and NCCRS have over 2,100 cooperating colleges that will consider accepting credits; the final decision is with the universities.
How much does Introduction to Artificial Intelligence cost and what is included?
$250, paid once, with lifetime access: all 69 video lessons, a practice quiz after each, the graded work and final exam, and an official transcript when you finish. Students who want several courses choose UPI Study PRO instead.
Which universities accept UPI Study credits?
UPI Study has signed partner agreements with universities such as Thomas Edison State University, University of the People and Charter Oak State College, and works with universities such as Penn State, SNHU and SUNY. ACE and NCCRS have over 2,100 cooperating colleges that will consider accepting credits; the final decision is with the universities. Use "Find my college" on this page to check yours.
How long does Introduction to Artificial Intelligence take?
It is self-paced with no deadlines. The 69 lessons total about 16 hours of video; most students add reading and practice time and finish in a few weeks to a few months, on their own schedule.
How are the exams done?
Quizzes and assignments are online in the course. The final exam is taken online with proctoring; a person on the UPI Study team checks assignments, usually within two days.
How do my credits reach my college?
When the course shows 100% complete, request the official transcript from your dashboard. It is sent to the college you name, and you can send it to more than one. Check your college first with "Find my college" on this page, or ask Athena.

Ready for the full course? Earn 3 credits in Introduction to Artificial Intelligence.

69 video lessons, a quiz after each one, and the graded work that earns the credit. Recommended for credit by NCCRS.

Enroll in Introduction to Artificial Intelligence →Download the syllabusEvery course with PRO →
Introduction to Artificial Intelligence
3 credits · NCCRSEnroll →