Skip to content

Module 04 — AI Literacy & Tools

Duration: 2–3 sessions Audience: Adults and older teens Prerequisite: Module 01, Module 02

The most-requested module across audiences. AI is changing job applications, customer service, education, and information ecosystems — Second Boot participants need to be on the using side, not just the receiving side.

Learning objectives

By the end of Module 04, a participant can:

  • Describe in plain language what a large language model does (and what it doesn't do)
  • Use a cloud AI tool (ChatGPT, Claude) productively and safely
  • Run a local AI model offline on their Second Boot device (if it has enough RAM)
  • Spot likely AI-generated text, images, or audio
  • Use AI tools for job search: cover letters, resume tightening, interview prep
  • Understand bias, hallucination, and what AI shouldn't be trusted with

Session breakdown

Session 1 — What AI actually is

  • "AI" vs. "machine learning" vs. "LLM" — the words people use
  • How an LLM works at a kitchen-table level (predicts next words from patterns)
  • What makes a model "good" — and what it still gets wrong
  • Hallucination: why models make things up, how to catch it
  • Bias: where it comes from, how it shows up
  • Demo: a working session with ChatGPT or Claude, including a deliberate hallucination caught in real time

Session 2 — Using AI productively

  • Prompting basics: be specific, give context, ask for format
  • AI for job search:
    • Tightening a resume
    • Drafting a cover letter from a job description
    • Practicing interview answers
    • Translating military, recovery, or unconventional experience into civilian-employer language
  • AI for writing: drafting, editing, summarizing
  • AI for learning: explaining a concept three different ways
  • What AI shouldn't be trusted with: medical diagnosis, legal decisions, current events, math without verification

Session 3 — Local AI offline (devices with 8GB+ RAM only)

  • Why local AI matters: privacy, no account needed, no internet needed
  • Install Ollama and pull a small model (Phi-3 Mini or Llama 3.2 3B)
  • Try a chat using Open WebUI
  • Compare: what's the same as ChatGPT, what's different
  • When local is the right choice (sensitive content, no connectivity, learning) — and when it isn't

Hands-on assignment

By the end of the module:

  1. The participant has had at least one substantive chat with a cloud AI tool
  2. They've drafted (or improved) one real thing they care about — a resume, a cover letter, a tough message, a study guide
  3. If their device has the RAM: Ollama is installed and they've had a chat offline
  4. They can name one thing AI got wrong during the module

Audience adaptations

Audience Emphasis
Recovery / reentry Job applications, recovery-friendly employer search, translating "gap years"
Incarcerated populations Heavy emphasis on local Ollama (no internet), educational use, supervised
Seniors Lighter touch — AI as a research assistant, not as a daily tool; scam-detection use
Veterans Translating MOS / military experience into civilian roles
Adults Practical productivity; awareness of AI in the workplace they're entering

Ethics and limits

The module explicitly covers what AI is not good for:

  • Medical or mental health decisions (refer to real clinicians)
  • Legal advice that matters (refer to real attorneys; for reentry participants, refer to fair-chance legal aid)
  • Current events without verification
  • Decisions about other people's lives
  • Anything where being wrong has real consequences and the answer isn't checkable

Assessment for award eligibility

A facilitator confirms (no written test):

  • Did the participant use AI to do one real thing they care about?
  • Can they describe one thing AI got wrong in their experience?
  • (Optional, RAM-dependent) Did they get a local model running?

If the first two are yes, Module 04 is complete.