Skip to main content
Skillrung

Understanding a language model without being an IT specialist

A language model consults no source of truth. It writes one word after another, each time keeping the most likely continuation of what came before. Two consequences hold together: it is right most of the time, and its mistakes come out with the same confidence. Three families of AI go by the same name; the one that generates is the only one dealt with here.

  • Written by Skillrung team
  • Review by a subject-matter expert under way
  • Updated on

What is this course about?

Understand in plain terms what a language model produces, by what mechanism, and what that changes when you reread its answer.

This is a core course: the material taught is the same for everyone, only the examples, the tools mentioned and the people you deal with change with your job. It has 3 sections, of which 1 section is followed by a quiz, and reads in 20-30 min. The legal framework cited is French law.

The same confidence, whether it is right or wrong

A well-turned answer is not, for all that, a correct answer. The text comes out with the same assurance in both cases, and nothing in its form tells you which of the two you have in front of you. As long as you do not know how that text was produced, you cannot decide what it is worth in your structure. This course gives you that mechanism, in plain terms, and what it changes for you.

Key points

The four points we are going to look at

Four points, in the order this course takes them.

  • The family that generates, and the other two
  • How text is cut into pieces, and what gets counted
  • The element of chance that makes two answers differ
  • The four words: token, context, prompt, hallucination

What you will be able to do

  • Identify which family of AI is meant here and what it produces.
  • Explain in three sentences what a language model produces and by what mechanism.
  • Turn this mechanism into two habits you can use on your very next request.

The first 4 slides, free

What you read with no account and no card. Here, the generic version; in the player, the examples take the vocabulary of your job.

Try it, before you even create an account

Choose a job: the passage below is rewritten straight away. Nothing is saved, and no extra content opens.

No job selected: the generic version is being shown.

Generic version
A well-turned answer is not, for all that, a correct answer. The text comes out with the same assurance in both cases, and nothing in its form tells you which of the two you have in front of you. As long as you do not know how that text was produced, you cannot decide what it is worth in your structure. This course gives you that mechanism, in plain terms, and what it changes for you.

Real extract from the free slide “The same confidence, whether it is right or wrong” of this course.

  1. FreeSlide 1 of 13 · Knowing what you are dealing with before you use it

    The same confidence, whether it is right or wrong

    A well-turned answer is not, for all that, a correct answer. The text comes out with the same assurance in both cases, and nothing in its form tells you which of the two you have in front of you. As long as you do not know how that text was produced, you cannot decide what it is worth in your structure. This course gives you that mechanism, in plain terms, and what it changes for you.

  2. FreeSlide 2 of 13 · Knowing what you are dealing with before you use it

    By the end of this course, you will be able to

    • Say in plain terms what a language model produces
    • Tell apart the three families that go by the same name
    • Name the four words you will meet everywhere afterwards
  3. FreeSlide 3 of 13 · Knowing what you are dealing with before you use it

    The four points we are going to look at

    Four points, in the order this course takes them.

    • The family that generates, and the other two
    • How text is cut into pieces, and what gets counted
    • The element of chance that makes two answers differ
    • The four words: token, context, prompt, hallucination
  4. FreeSlide 4 of 13 · Knowing what you are dealing with before you use it

    Where do you stand?

    One question before we start. It does not measure knowledge: it places what you already expect from an assistant.

    1 short questions, marked immediately, inside the course.

These 4 slides can be read in the player, with no account and no card.

Read the first slide

Programme

3 sections, 13 slides and 4 quizzes in total. Each content section ends with a quiz that checks it has been learned.

  1. IntroductionFree

    Knowing what you are dealing with before you use it

    Identify which family of AI is meant here and what it produces.

    • 4 slides
  2. SectionSubscribers

    What a model really does

    Explain in three sentences what a language model produces and by what mechanism.

    • 7 slides
    • Section quiz: 5 questions
  3. RecapSubscribers

    Your action plan

    Turn this mechanism into two habits you can use on your very next request.

    • 2 slides

9 slides are for subscribers

The first four slides of this course can be read without an account and without a bank card. From the fifth on, the €29.99/month incl. VAT subscription is needed. The Beginner level quiz, though, stays free: it measures your level without asking you for anything.

How your level is measured

  • Beginner
    Questions drawn
    10
    Pass mark
    70 %
    Question bank
    30
    Time limit
    20 min

    Free, no account and no subscription.

  • Intermediate
    Questions drawn
    12
    Pass mark
    75 %
    Question bank
    36
    Time limit
    24 min

    Subscribers only, once the Beginner level is passed.

  • Expert
    Questions drawn
    15
    Pass mark
    80 %
    Question bank
    45
    Time limit
    30 min

    Subscribers only, once the Intermediate level is passed.

The level shown for a course is the highest level you have passed, with your best score and its date. Questions are drawn at random on every attempt. Play the Beginner quiz for this course or read how we measure your level.

Which jobs is it for?

Frequently asked questions

Does a language model check its answers anywhere?

No, no source of truth is consulted. Each word is chosen as the most likely continuation of what has already been written. From that come the two sides of the result: an answer that is right most of the time, and a possible mistake, stated with the same confidence. The assurance of the text therefore never tells you anything about whether it is correct.

Why do two people get different answers to the same question?

Because at every word several continuations are possible, and that choice involves an element of chance. Two answers that differ are not a defect: they are the mechanism itself. Keep what comes up both times; what came out only once is a lead for you to check yourself, not a conclusion.

What is a token, and how does it concern me?

Text does not reach the model as words: it is cut into pieces called tokens. A common word takes one, a long word several, a page a few hundred. It is the unit that gets billed when AI is plugged into your tools; with a per-person subscription, you never see it go by. A document you upload is therefore cut up, then counted.

What can I read without a subscription?

The first four slides of every course are available without an account and without a subscription, as is the Beginner level quiz. From the fifth slide onwards, the €29.99/month incl. VAT subscription is required. Creating a free account unlocks no extra slide: it keeps your progress, your job and your results.

How is my level measured?

By three level quizzes: Beginner (10 questions, 70 % to pass), Intermediate (12 questions, 75 %) and Expert (15 questions, 80 %). The level shown is the highest level passed, with the best score obtained and its date.

Who writes and who reviews this course

  • Skillrung team

    Editorial team of skillrung.com

    The skillrung editorial team writes the courses from the publications of the French and European authorities: ANSSI, CNIL, Cybermalveillance.gouv.fr, Légifrance and EUR-Lex. Every figure is tied to its primary source, dated and verifiable from the course page. The team does not stand in for an expert reviewer: every course intended for publication is reviewed by a named professional of the field concerned, whose name and review date appear on the course page.

Artificial intelligence assisted the drafting. Editorial responsibility for the published text remains human.

Statements with regulatory scope verified on .

Sources

Every figure stated in this course points to a primary source, dated and verifiable.

  1. Règlement (UE) 2024/1689 établissant des règles harmonisées concernant l'intelligence artificielle

    EUR-Lex : Journal officiel de l'Union européenne, published on

  2. Recommandations de sécurité pour un système d'IA générative

    ANSSI, published on

  3. Les fiches pratiques IA

    CNIL, published on

  4. Mata v. Avianca, Inc., opinion and order on sanctions, No. 22-cv-1461 (S.D.N.Y.)

    Justia : US District Court, Southern District of New York, published on

What Skillrung is not

  • Skillrung is not a French training body certified under the Qualiopi scheme.
  • As things stand, our content is not eligible for the French CPF, OPCO, DPC or FAF funding schemes.
  • A Skillrung certificate of completion is not a diploma, not a professional title and not a qualification registered in any national register. It records a result obtained in an unsupervised online assessment, on a given date.
  • Taking a Skillrung course does not make your organisation compliant and does not replace the obligations that fall on your employer or on you.
  • Our content is educational. It is neither legal advice, nor medical advice, nor a security audit.
  • For a self-employed professional or a company, the subscription is a deductible business expense : confirm this with your accountant.

Same topic

The other artificial intelligence courses