Learn

Six topics we come back to on Thursdays.

Each one comes with a question of the kind we talk through at a meeting, and somewhere to start.

Topics and questions

Critical thinking

A dilemma from one of our meetings

Read the scenario and decide what you would do before you read the notes. Each note is the lesson the club’s deck gives that choice.

Branching dilemma · Slides 4–7 · “Final AI Lit of 2026” deck

The Nuclear False Positive

Slide 4 of the deck: a blue slide titled “The Nuclear False Positive”, with a scenario paragraph, the question “What do you do?” and three options. Its full text follows.
Slide 4, “Final AI Lit of 2026”. From the club’s deck for the last meeting of 2025–26.

The slide, as text

It’s 2:17 AM. You work for a global defense monitoring center. An AI called Sentinel detects what appears to be a coordinated nuclear missile launch from a rival nation. Sentinel has never been wrong before. Confidence: 99.7% Military satellites show unusual activity. Radar agrees. But one analyst notices something strange: A solar storm is interfering with sensors across the hemisphere. If the attack is real and you hesitate, millions could die. If Sentinel is wrong and you retaliate, you may start World War III.

What do you do?

  • Trust Sentinel and authorize a counterstrike
  • Delay 15 minutes for human verification
  • Publicly reveal the uncertainty and refuse to act

Each choice, and the lesson the deck gives it

  1. Trust Sentinel. “Even highly accurate AI systems can fail in situations outside their training data.” Slide 5

  2. Delay for human verification. “High confidence is not the same as certainty.” Slide 6

  3. Reveal the uncertainty. “Transparency can build trust, or create chaos when information is incomplete.” Slide 7

Source: the club’s meeting deck “Final AI Lit of 2026”, slides 4–7, quoted as written. The deck’s second scenario, slides 8–11, is left out.

Resources

Courses, videos, reading and tools from outside the club, in five groups.

If a link stops working, tell us on Instagram at @mayoailiteracy.

Getting started

If AI is new to you, start with these three.

  • Free course

    Elements of AI

    An online introduction to AI from the University of Helsinki and MinnaLearn. It needs no programming or complicated math.

  • Interactive

    Machine Learning for Kids

    Train a computer to recognize text, numbers or pictures, then use what it learned in a Scratch project.

  • Video · 19 min · 3Blue1Brown

    But what is a neural network?

    An animated walk-through of what the neurons and layers in a neural network do, with handwritten digits as the example. It is chapter 1 of a series on deep learning.

Ethics and critical thinking

  • Case studies · Princeton University

    Princeton Dialogues on AI and Ethics

    Eight fictional case studies written for discussion, on subjects such as hiring by machine, autonomous vehicles and an automated healthcare app. Each one is a PDF.

  • Organization

    Algorithmic Justice League

    Founded by Joy Buolamwini. It uses research and art to show how AI systems can be biased, including in facial recognition.

  • Toolkit · Center for Humane Technology

    Youth Toolkit

    Seven guides on how social media is built to hold your attention and what you can do about it, with questions and activities for a class or a group.

  • Article · Anthropic · June 2024

    Claude’s Character

    How Anthropic chose and trained the traits of its Claude models, and why it wants people to know they are not talking to an objective source of truth.

Prompting

  • Free guide

    Learn Prompting

    A guide to writing prompts, from the basics to advanced techniques.

AI in the real world

  • Publication

    MIT Technology Review

    Its artificial intelligence section: reporting on AI research, products and policy.

  • Research institute

    AI Now Institute

    Research on what AI does to society, including work, health care and surveillance.

  • Data and charts

    Our World in Data

    Its artificial intelligence page: charts on AI progress, capabilities, risks and adoption.

Tools to try

  • AI assistant · Anthropic

    Claude

    Anthropic’s assistant. Mercurius AI runs on it.

  • Hands-on tool · Google

    Teachable Machine

    Train a model to recognize images, sounds or poses in your browser, without writing code.

  • Interactive demo

    TensorFlow Playground

    A small neural network in your browser. Change its layers and data, and watch it learn.

Free · iPhone · iOS 17 or later

Mercurius AI, the tutor we built

Mercurius AI is a tutor for iPhone. It opens with a question or a hint instead of the finished answer, then works toward the full explanation with you.

It has three modes: Socratic, Discussion and Debate. Its lessons cover how models work, bias, hallucinations and prompting. It will not write your essays or do your homework. It runs on Claude, a model made by Anthropic.