UPI Study › Courses › Introduction to Artificial Intelligence › Free lessons › Lesson 6.1 Free with a free account
Analytical Thinking and the Role of Mathematical Logic
Introduction to Artificial Intelligence · Chapter 6: Logic, Reasoning, and Knowledge SystemsSection 6.1 · lesson 39 of 6913 min · free with a free account
Free preview
Create a free account to keep watching
It takes a minute and costs nothing. Your free account unlocks all of Chapter 1 and 3 more lessons of Introduction to Artificial Intelligence.
Lesson complete
Earn the 3 college credits for Introduction to Artificial Intelligence
Enrolling adds a quiz after each lesson, the graded work and the official transcript. Recommended for credit by NCCRS.
Next lesson: Logic and Computation in Intelligent Systems →← Sequence-Aware Learning with Recurrent NetworksNext lesson: Logic and Computation in Intelligent Systems →
In this lesson
How arguments, validity and fallacies ground the logic AI systems use.
- Distinguish deductive, inductive and abductive reasoning by direction and certainty
- Explain the difference between valid and sound arguments
- Test an argument form for validity by searching for a counterexample
- Identify common formal and informal fallacies
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 (1,780 words). Read along, or search it.
Opening Question
Three Arguments, Unequal Trust
Read three pieces of reasoning. Everyone who subscribed receives the newsletter. Priya subscribed. So Priya receives the newsletter. Every swan I've seen is white, so all swans are white. And: the grass is wet, so it probably rained. All three move from reasons to a conclusion. Yet they aren't equally trustworthy. The first can't go wrong if its starting points are true. The second could fall to one black swan. The third might be a sprinkler. An AI system that works with knowledge must know which kind of conclusion it has reached. That's the job of logic: the study of correct reasoning.
Definition
Statements, Premises and Conclusions
Analytical thinking means breaking reasoning into its parts, and checking how well they fit. The basic unit is a statement: a sentence that's either true or false. An argument is a set of statements in which some, the premises, are offered as reasons for another, the conclusion. In the newsletter case, the premises are: everyone who subscribed receives it, and Priya subscribed. The conclusion: Priya receives it. Premise words include because, since, and given that. Conclusion words include therefore, so, thus, and hence. Then two questions drive everything: does the conclusion follow? And are the premises true?
Three Kinds
Deduction, Induction and Abduction
Reasoning comes in three main kinds, which differ in direction, and in how certain the conclusion is. Deductive reasoning goes from general rules to a specific case, and it's certain if the premises are true. All mammals breathe air; a dolphin is a mammal; so a dolphin breathes air. Rule-based systems work this way. Inductive reasoning goes from specific cases to a general rule. It's probable, stronger with more evidence, but never certain. The last 50 deliveries arrived on time, so the next one probably will. That's machine learning. And abductive reasoning goes from an observation to the best explanation. It's plausible, but other explanations may exist. The street is wet; the best explanation is rain. Diagnosis and fault finding work this way.
Classic Example
The Bag of White Beans
A classic example uses a bag of beans, and three statements. A rule: all the beans in this bag are white. A case: these beans came from this bag. And a result: these beans are white. Each kind of reasoning takes two of the statements as given, and infers the third. Deduction takes the rule and the case, and infers the result. Induction takes the case and the result, and infers the rule. And abduction takes the rule and the result, and infers the case: these beans probably came from this bag.
Trade-Off
Certainty Versus New Knowledge
There's a trade-off. A deductive conclusion contains nothing that wasn't already hidden in the premises, which is why it can be certain. Induction and abduction go beyond the premises. They create new knowledge, like a general law or a diagnosis, but they can be wrong. For centuries, people in Europe believed all swans were white. In the late 1600s, black swans were found in Australia, and the law collapsed. Each white swan adds a little support, but one contrary case ends it. So each kind is judged differently. A deductive argument is valid or invalid, all or nothing. An inductive one is strong or weak, by degrees. And an abductive one, by how good the explanation is, compared with the alternatives.
Quick Check
A Brake Squeal Calls for Abduction
Quick check. A mechanic hears a high squeal whenever a car brakes, and concludes that the brake pads are probably worn. Which kind of reasoning is that? Pause, and decide. Abductive. It starts from an observation, the squeal, and reaches the best explanation, worn pads. Other explanations could exist, like a stone caught in the brake. It isn't deductive, because no rule makes worn pads certain. And it isn't inductive: it doesn't generalize a rule from many cases. It explains one observation.
Key Terms
Valid Versus Sound Arguments
For deductive arguments, two similar words mean different things. Valid: it's impossible for all the premises to be true while the conclusion is false. Validity is about form, not about what's actually true. Sound: valid, with all premises actually true. A sound argument must have a true conclusion. All mammals are warm-blooded; whales are mammals; so whales are warm-blooded. Valid, with true premises: sound. All birds can fly; penguins are birds; so penguins can fly. Valid, but the first premise is false, so not sound. A valid argument can reach a false conclusion. And: if it's raining, the street is wet; the street is wet; so it's raining. That one isn't even valid. The street may have been washed.
Mechanical Check
Hunting for a Counterexample Row
Because validity depends only on form, a machine can check it, without understanding the sentences. Replace each simple statement with a letter, and list every combination of true and false. The argument is invalid if some row makes all the premises true, and the conclusion false. That row is a counterexample. Form one: if P then Q; P; therefore Q. In the only row where both premises are true, Q is true too. No counterexample, so form one is valid. Form two: if P then Q; Q; therefore P. Now look at the row where P is false and Q is true. Both premises hold, but the conclusion fails: a counterexample. The street is wet, yet it isn't raining.
Formal Fallacies
Affirming the Consequent, Denying the Antecedent
A fallacy is a flaw in reasoning that makes an argument look stronger than it is. Some are errors of form, which that check would catch. Affirming the consequent is form two. If a message is spam, it contains a link. This one contains a link, so it's spam. But genuine messages contain links too. Denying the antecedent: if a message is spam, it contains a link. This one has no link, so it isn't spam. But spam can arrive without a link.
Informal Fallacies
False Cause, Hasty Generalization and More
Other fallacies are informal: the argument sounds acceptable, but its content or context undermines it. Here are five, with examples from AI work. Hasty generalization: the model got 9 of 10 test emails right, so it's 90 percent accurate on all mail. False cause: sales rose after the chatbot launched, so the chatbot caused the rise. Improper appeal to authority: a famous actor says this medical-advice bot is reliable, so it must be. And an attack on the person: that bug report came from a new intern, so ignore it. False dilemma: either we deploy the model today, or the project is a failure, as if no other options existed.
More Fallacies
Straw Man, Circular Reasoning, Slippery Slope
Four more. A straw man distorts a position to make it easy to attack: she wants more testing, so she must want to delay every release forever. Circular reasoning hides the conclusion in the premises: the model is reliable, because it always gives correct answers. Appeal to popularity: everyone uses this library, so its results must be correct. And a slippery slope claims, without evidence, that one step leads to disaster: allow one exception, and soon every rule will be ignored. Remember, a fallacious argument can still have a true conclusion. The argument just hasn't shown it. So ask three questions: what are the premises, does the conclusion follow, and are the premises true?
Quick Check
Crosswalk Paint and False Cause
Quick check. A city repainted its crosswalks in May, and in June, pedestrian injuries fell. So the city says the new paint cut injuries. Which fallacy does that risk? Pause, and decide. False cause. Injuries fell after the repaint, but after isn't because. A new speed limit, a rainy June, or fewer walkers could explain it. A straw man distorts someone's position, and a false dilemma offers too few options. Neither fits. Here, one event following another is taken as proof that the first caused the second.
Why Formal Logic
Precision, Checking and Explanation
Everyday language is rich, but it's a poor medium for a machine that must reason reliably. Formal logic writes statements in a precise artificial language, with exact rules for drawing conclusions. That gives precision: every statement has exactly one meaning. Mechanical checking: a program can test an argument without understanding it, like the truth check. Guaranteed conclusions: deduction from true premises gives true conclusions. Explanation: the chain of steps can be shown, like a system citing the exact rules behind its decision. And sharing: knowledge written formally can be combined and reused by different programs, like a shared body of facts about drugs and their interactions.
Limitations
Exceptions, Uncertainty and Common Sense
Formal logic also struggles. Exceptions: a rule like birds fly fails for penguins, and each exception must be written in. Rules that can be revised, or probabilities, help. Uncertainty: plain logic knows true and false, not probably. Probabilistic models help. Common sense: everyday knowledge is enormous, and hard to write down. And logic doesn't discover new rules from examples by itself. So modern AI combines approaches. Learning copes with uncertainty and scale, and logic supplies precision and checkable reasoning, where mistakes are costly.
Comparison
Informal Versus Formal Reasoning
Informal reasoning is what people do every day, in ordinary language, where meaning depends on context, tone, and shared knowledge. Ambiguity is common, and checking takes judgment and discussion. Formal reasoning uses precisely defined symbols, with meaning fixed by definition. Ambiguity isn't allowed, and checking is mechanical. It's less flexible, slower, and the risk moves to getting the translation wrong. Translating forces ambiguity into the open. Every student passed an exam can mean one particular exam that everyone passed, or that each student passed some exam, possibly a different one. A formal language makes the writer choose.
Step by Step
Four Steps to Formalize an Argument
To formalize an argument, use four steps. Take: if the sensor reading is above 80, the alarm sounds. The reading is 90. So the alarm sounds. First, find the simple statements: the reading is above 80, and the alarm sounds. Second, name them: P and Q. Third, connect them: if P then Q, and P. Fourth, check the form: if P then Q, P, therefore Q. That's form one, which is valid. A machine can confirm the validity. But whether the premises are true of the real sensor, still has to be checked against the world.
Key Takeaways
Arguments, Validity and Logic in AI
To sum up. An argument has premises and a conclusion. Always ask whether the conclusion follows, and whether the premises are true. Deduction is certain if the premises are true. Induction is probable, and abduction plausible: they add knowledge, but can be wrong. A valid argument can't have true premises and a false conclusion, and a sound one is valid with true premises. Validity can be checked mechanically, by hunting for a counterexample. And AI uses formal logic for precision and explanation, combined with learning, for exceptions and uncertainty.
Think About It
Spotting Fallacies in Everyday Claims
Something to think about. Find a fallacy in an ad, a headline, or a social media post. Which fallacy is it, and what would the author need to show for the argument to work? And when a system gives a convincing explanation for a machine fault, which kind of reasoning is that? How could you find out whether it's right?
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
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 UniversitySigned partner
University of the PeopleSigned partner
Charter Oak State CollegeSigned partner
Strayer UniversitySigned partner
Columbia Southern UniversitySigned partner
Indiana TechSigned partner
Penn StateWorks with NCCRS
Southern New Hampshire UniversityWorks with NCCRS
SUNYWorks 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 Analytical Thinking and the Role of Mathematical Logic
What is the difference between deductive, inductive and abductive reasoning?
What is the difference between a valid and a sound argument?
How do you check whether an argument form is valid?
What is the difference between a formal and an informal fallacy?
About the Introduction to Artificial Intelligence course
How many college credits is Introduction to Artificial Intelligence?
How much does Introduction to Artificial Intelligence cost and what is included?
Which universities accept UPI Study credits?
How long does Introduction to Artificial Intelligence take?
How are the exams done?
How do my credits reach my college?
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.
3 credits · NCCRSEnroll →