← ExitAI Fluency Foundations
0 / 14 lessons0 XP
Lesson2of 14

0of15read0 XP

AI Fluency FoundationsLesson 2.2

How AI learns: the machine, no math

Explain in plain words what an AI language model does: it predicts the next likely piece of text from patterns it learned, rather than looking up facts in a database[^1].

Time
20–25 min
Type
exercise
Bloom
Apply → Create
XP
100
Concept architecture for How AI learns: the machine, no math

Architecture diagram for How AI learns: the machine, no math. A blindfolded improv actor on a stage, mid-line, with speech flowing out confidently. Behind them, instead of a reference library, a stream of overlapping scripts they've absorbed. One branch of the speech is labeled 'fluent and right' and another, identical-looking branch labeled 'fluent and invented (hallucination)', to show both come from the same source. Caption: 'It predicts the next likely line. It can't tell true from plausible.' Warm gold on near-black.

Lesson 2.2 — concept architecture

You'll be able to

  • Explain in plain words what an AI language model does: it predicts the next likely piece of text from patterns it learned, rather than looking up facts in a database[^1].
  • Explain why the same machinery that makes AI fluent is exactly why it confidently makes things up, so hallucination is a feature of how it works, not a bug bolted on[^2][^3].
  • Explain how bias enters through training data: whatever patterns the model learned from, it repeats.

Key concepts · tap to reveal

1/15·Watch·Beat 1 · Hook

0%

Hook

Here's why it's brilliant and why it lies, the machine behind every move you're about to learn. A chatbot promises two-day shipping on a six-week order. A coding tool invents a function. Same machinery makes AI useful and makes it make things up. Know it, and nothing surprises you.

Prompt Labruns here · claude

Your task  Write a prompt that asks Claude to recommend the right AI setup for a real task you're facing — then weigh its answer against this lesson, "How AI learns: the machine, no math."

a strong prompt:role · context · task · format · example

⌘↵ to run
A blindfolded improv actor on a stage, mid-line, with speech flowing out confidently. Behind them, instead of a reference library, a stream of overlapping scripts they've absorbed. One branch of the speech is labeled 'fluent and right' and another, i
Diagram · generated brief

Exercise · scenario

You ask an AI tool for the exact population of a real but obscure town, and it gives you a precise-looking number with total confidence.

Deliverable

Add a page to your **AI Fluency Playbook** titled **How the machine works (in my own words)**. Write three short things:

Reveal model answer

High invention risk

Practice · Scenarios

0 of 2 revealed

Scenario 1 of 2

An AI tool answers a medical question with confident, fluent prose and lists three journal citations that look real, but none of the papers exist when you search for them.

Step 1 · Classify

Sources

  1. [1]Google Cloud Generative AI Leader Exam Guide·Google Cloud Generative AI Leader Exam Guide > Task 3.3: Identify grounding techniques and their use cases. Considerations include: (2025) · Vendor
  2. [2]AWS Certified AI Practitioner (AIF-C01) Exam Guide·AWS Certified AI Practitioner (AIF-C01) Exam Guide > Domain 2: Fundamentals of Generative AI > Task 2.2: Understand the capabilities and lim (2025) · Vendor
  3. [3]arXiv API·arXiv API > Knowledge Graphs, Large Language Models, and Hallucinations: An NLP Perspective > Abstract (2025) · Research
Capstone artifact · auto-graded

Submit your work for review

Paste your capstone artifact below. You'll get back a 4-level rubric grade, per-criterion feedback, and three concrete edits to strengthen it.

0 chars · minimum 50