무료

수료증 무료 발급 — 2027년 3월 31일까지.

PT737E Understanding AI for Ministry

Lecture 1-1: What the Word "AI" Points To

목차 31
PT737E Understanding AI for Ministry
제 0 장 · Introduction: A Tool That Continues Words
  1. Introduction: A Tool That Continues Words
제 1 장 · Week 1: What AI Is
  1. Lecture 1-1: What the Word "AI" Points To
  2. Lecture 1-2: How a Language Model Makes Words
  3. Lecture 1-3: Sounding Right Is Not Knowing
  4. Lecture 1-4: Reading a Conversation Closely
  5. Lecture 1-5: Practice Lab, Read an AI Answer
제 2 장 · Week 2: Limits and Errors
  1. Lecture 2-1: Invented Facts
  2. Lecture 2-2: Old Knowledge, Missing Sources, Uneven Coverage
  3. Lecture 2-3: Scripture and Theology, Where to Be Most Careful
  4. Lecture 2-4: Building a Checking Plan
  5. Lecture 2-5: Practice Lab, Find the Errors
제 3 장 · Week 3: Asking, Checking, Protecting
  1. Lecture 3-1: Asking Well
  2. Lecture 3-2: Checking What You Are Given
  3. Lecture 3-3: What Must Never Be Typed In
  4. Lecture 3-4: Keeping Notes and Giving Credit
  5. Lecture 3-5: Practice Lab, Mend the Prompts, Mark the Secrets
제 4 장 · Week 4: Ministry Work: Administration and Teaching
  1. Lecture 4-1: Hand It Over, Help With It, Keep It
  2. Lecture 4-2: Office Work and Communication
  3. Lecture 4-3: Teaching and Study Materials
  4. Lecture 4-4: Where Teachers and Leaders Draw the Line
  5. Lecture 4-5: Practice Lab: Hand Over, Help, or Keep
제 5 장 · Week 5: Preaching and Care
  1. Lecture 5-1: Preaching and AI: A Theological Question
  2. Lecture 5-2: Care and Counselling
  3. Lecture 5-3: When Members and Young People Use AI
  4. Lecture 5-4: The Minister's Own Life
  5. Lecture 5-5: Practice Lab: Judge the Scene, Write a Position Paragraph
제 6 장 · Week 6: Staying Responsible for the Long Run
  1. Lecture 6-1: The Learner's Honesty
  2. Lecture 6-2: A Community's Policy
  3. Lecture 6-3: Technology That Keeps Changing
  4. Lecture 6-4: Writing Your Own Rule
  5. Lecture 6-5: Practice Lab, My Personal Guidelines

핵심만 먼저 보기 ↓

19분

Ask ten people what "AI" means and you will hear ten answers. One thinks of a robot in a film. One thinks of the voice in a phone. One thinks of a machine that will take her job. One thinks of a program that wrote a letter for him last week. A deacon in charge of the parish office may think of the spam filter that has quietly sorted the mail for years and never been called "AI" at all.

The word is a label, and a label can be stuck on very different things. A minister who wants to decide whether and how to use such tools needs first to know what is being pointed at. This lecture sorts the word into three families, shows where each already sits in ordinary life, and then gives you a short set of questions for reading the claims, both the excited ones and the frightened ones, that come with the word.

The course returns, in every chapter, to one claim:

An AI system is a tool that continues language plausibly; it is not a knower. So a minister learns how it works, checks what it says, keeps the human work (proclaiming the Word, caring for people, keeping confidences) in human hands, and is honest about what was used.

This lecture does not say whether you should use these tools. It says what they are, in terms a non-specialist can hold. No programming and no mathematics are needed anywhere in this course.

이 19분으로 얻는 것

  • Describe three families of systems that are called AI: rule-following programs, programs trained on data, and generative systems.
  • Name ordinary places where the first two families already work in daily life.
  • Say what makes a generative system different from a program that only sorts or predicts.
  • Ask five questions that turn a large claim about AI into something that can be examined.

A simple man believes everything, but the prudent man carefully considers his ways." — Proverbs 14:15

World English Bible

Test all things, and hold firmly that which is good." — 1 Thessalonians 5:21

WEB

The first verse warns against believing everything and commends the person who thinks about where a path leads. The second tells believers to test what they are told. A course on AI does not need a special rule for these tools. It needs the old habit of careful hearing, applied to a new kind of speaker.

1. A label for many things

The term "artificial intelligence" was coined as a name for a field of research, and since then it has been used for an enormous range of programs. Some of them are old and plain. Some are new and striking. Some are marketing words. The label does not tell you how a thing works, how reliable it is, or what it is for.

For a minister this matters because the question "should we use AI?" has no single answer. It is like asking "should we use paper?" Paper may be a prayer card, a legal form, or a wrapper for fish. The useful question is always about one particular tool doing one particular job.

So we begin by sorting. The sorting below is a teaching map, not a technical standard. Specialists divide the field in other ways, and real systems often combine families.

2. Family one: programs that follow rules

The oldest family is a program in which a person has written down the rules in advance. If the date on a form is typed in the wrong shape, show a message. If the balance is below a limit, send a warning. If a name is on the list, print a badge.

Such a program is predictable. Given the same input it gives the same output, and the person who wrote the rules can say why. Its weakness is the opposite: it does nothing that nobody thought of. A rule-following program meets a case its author never imagined and either stops or does something foolish.

Many things that are described as "smart" in advertising are programs of this kind. That is not an insult. A reliable set of rules is often exactly what an office needs.

3. Family two: programs trained on data

The second family does not have rules written by hand. Instead, a system is shown a very large number of examples and adjusts itself until it can find the patterns in them. After this training, it can be given a new case and give a result based on those patterns.

(An invented example.) A parish office receives hundreds of emails each week. A mail program has been shown a great many messages that people marked as unwanted, and a great many that they kept. It now sorts new messages with fair success. No one wrote a rule saying "reject anything with this word." The program found a pattern across the examples that no person listed.

Other places where you may already meet this family: turning speech into written words, suggesting the next word as you type on a phone, grouping photographs by what they seem to show, offering a route that avoids slow roads, translating a short sentence. These are usually described as doing one thing, and doing it with a rate of error that is acceptable for that thing.

The weakness of the family is that the system is only as good as its examples. If the examples did not include a case, or included it badly, the result can be wrong in ways that are hard to predict, and the system cannot explain itself in the way a person following written rules could.

4. Family three: generative systems

The third family is the one that has drawn most attention. A generative system does not only sort or predict. It produces something new in response to a request: a paragraph, a picture, a spoken voice, a piece of computer code.

The generative systems that matter most for this course are language models, programs trained on a vast body of written text that can continue a piece of writing in a way that reads as natural. People usually meet them as a chat assistant: you type a request and it replies in fluent sentences, and you may then ask a follow-up. Lecture 1-2 explains in plain terms how such a system makes its words. For now, hold on to two facts.

First, it produces text; it does not retrieve a stored answer in the way a library or a dictionary does. Second, because it is built to produce text that reads well, the text reads well whether or not it is correct. Everything else in this course follows from those two facts.

Image generators, voice generators and tools that make video belong in the same family. They produce things that look or sound real. They raise questions of their own, among them the question of what is real, but the same habit of testing applies to them.

The one thing to carry forward

Rule-following programs do what they were told. Trained programs find patterns in examples. Generative systems produce new material that fits the patterns. Only the last of these writes sentences that sound like a person speaking.

5. Where these already sit in daily life

You need not search far. Mail sorting, route suggestions, phone keyboards that guess, captions on a video, a bank's warning about an unusual payment, the order in which a list of items is shown to you: most of these use the first two families, and for years. The newer element is the third family, which writes and speaks.

It helps to notice this because it takes the drama out of the word. A congregation member who says "my church uses AI" may mean a spam filter. A pastor who says "we do not touch AI" may be using three tools that fall under the label already. When your community writes a rule (we will come to this in Week 6), it should say what it means: "chat assistants that write text," not "AI."

6. Narrow tools

Another thing to know is that nearly everything you will meet is built to do a particular kind of job. A system that translates is not a system that plans a budget. A chat assistant is built to continue text, and it can be asked to do many text jobs, but that does not make it a general mind. When an advertisement speaks of a tool that "can do anything," the right response is a question: do what, and how do you know it is done correctly?

This course says nothing about what future systems may or may not become. Predictions of that kind change often, and a lecture that rests on one would soon be out of date. It describes the kind of tool that a minister can pick up now and gives questions that will still be useful when the tools change.

7. Reading the claims

Claims about AI come in two moods. The excited mood says that a system will transform everything, replace whole professions, or solve problems nobody could solve. The frightened mood says that a system will destroy jobs, deceive everyone, or turn against its makers. Both moods are loud. Neither is evidence.

Five questions turn a large claim into something small enough to examine.

  1. What exactly does it do? Not "uses AI" but "sorts these messages," or "writes first drafts of notices."
  2. What was it trained on, and what was left out? The answer is often not known to the user. That in itself is useful to know.
  3. Who is making the claim, and what do they gain? A seller of a tool, a writer who needs readers, and a researcher who wants a grant all have an interest. That does not make them wrong, but it tells you to look for a second voice.
  4. How was it tested, and on whom? A tool tested on one kind of text may fail on yours.
  5. What happens if it is wrong? A wrong route costs ten minutes. A wrong sentence in a letter to a grieving family costs much more. The cost of an error is the reason to check, and it is under your control even when the tool is not.

(An invented example.) A volunteer coordinator reads a sentence in a newsletter: "AI will soon run every church office." She asks the five questions. What does it do? The article does not say. Trained on what? Unknown. Who gains? The article is an advertisement for a course. Tested on whom? No test is given. If it is wrong? She would lose nothing by waiting. She decides to keep doing the roster by hand and to ask her office team which tasks cost them the most time.

Notice that she did not decide the tool was bad. She decided that the claim had not told her anything yet.

8. Fear and enthusiasm

A minister may feel either pull, and both are understandable. Some leaders are excited by what they see a chat assistant do in a minute. Some are uneasy because they have heard of misuse. Neither feeling is a mistake. The mistake is to let the feeling stand in for an examination.

There is a further set of serious questions that this course does not argue: who is responsible when a system causes harm, how bias in data affects people, what surveillance and the loss of work do to communities. Those belong to a separate course, Christian Ethics of AI (STEL5E). Here, bias will appear only as a limit that can show in what a tool produces. Likewise the question of what it means to be a human person, beside a machine that writes like one, is taken up in AI and Human Understanding (STEL6E), and the wider theology of technology in STEL7E.


Diagram 1. Three families behind one word

Three Families Behind the Word AI Different things carry one label Family Rule-following programs Who decides: A person writes the rules in advance Example: A form that refuses a date typed in the wrong shape Weakness: Does only what its rules cover; fails on the unexpected Programs trained on data Who decides: Patterns are found in many examples Example: A filter that sorts mail into wanted and unwanted Weakness: Is only as good as its examples; can be wrong in a new case Generative systems Who decides: Patterns are used to produce new text, images or sound Example: A chat assistant that writes a draft when asked Weakness: Can produce fluent material that is wrong Ask of any claim that says AI: which family is this, and what exactly does it do?

Diagram 2. The five questions at a glance

Question What it protects you from
What exactly does it do? A large word hiding a small or unknown job
What was it trained on? Assuming it knows your community or your tradition
Who gains from the claim? Taking an advertisement for a finding
How was it tested, and on whom? Trusting a result from a different kind of case
What happens if it is wrong? Using a tool where an error would do real harm

핵심 요약

  • "AI" is a label stuck on very different things. Ask what exactly a tool does before asking whether to use it.
  • Three families help: rule-following programs, programs trained on data, and generative systems that produce new material.
  • The first two families already work in mail sorting, route suggestions, captions and similar tools. The newer element is the third family, which writes and speaks.
  • A generative language model produces text that reads well whether or not it is correct.
  • Most tools are narrow. A claim that a tool "can do anything" should be answered with: do what, and how do you know?
  • Five questions turn a claim into something examinable: what does it do, trained on what, who gains, tested on whom, what if wrong.
  • Ethics, personhood and the theology of technology are taught in STEL5E, STEL6E and STEL7E.
Front Back
Three families called AI Rule-following programs; programs trained on data; generative systems.
Training Showing a system a very large number of examples until it finds patterns it can use on new cases.
Generative system One that produces new text, images, sound or code in answer to a request.
Why fluency is not accuracy The system is built to make text that reads well, and it does so whether or not the text is correct.
The five questions What does it do? Trained on what? Who gains? Tested on whom? What if it is wrong?

One sentence to remember Before asking whether to use "AI," ask what exactly the tool does, and what it costs if it is wrong.

This week's action Write down three tools you meet in a normal week, and put each into one of the three families. Do not use any AI tool to do this.

이 과목 요약 전체 보기 →

스스로 확인해 보세요

읽었는지가 아니라 설명할 수 있는지를 묻는 것입니다. 체크는 이 기기에만 저장됩니다.

묵상 및 토론

  • List three tools you or your community already use. Which of the three families is each in?
  • Which of the two moods, excited or frightened, do you meet more often among the people you serve? What does each mood leave out?
  • Take one claim about AI you have recently heard or read. Put the five questions to it. Which could you answer, and which could not be answered at all?
  • Proverbs 14:15 sets the simple person who believes everything beside the prudent one who thinks. What would it look like to be prudent about a tool, not about a person?
  • Why might a community rule that says "no AI" be hard to follow, and what could be said in its place?

로그인하시면 여기에 답을 적고 보관하실 수 있습니다

다음 단계

여기까지가 강의입니다. 아래는 읽은 사람들이 남긴 글입니다.