AI English ShiftLEARN IT. USE IT.日本語한국어
AI LITERACY × ENGLISH LITERACY

Understand AI.
Make it your English.

AI literacy. English literacy. Built together.

Start the first lesson — freeSee how it works
Free to learn · No signup required
Mina reads about AI on her laptop.
01Read it
How does RAG work?
One idea. Two skills.AI-generated illustration

You’ve saved the link. Now what?

Too much to follow.

The concepts don’t connect yet.

It makes sense—until it’s your turn to explain.

Three parts. One learning journey.

We curate original sources, build the foundations, and give you a place to practise.

Courses build the foundations

Read with translations. Select a phrase. See how the concepts connect.

Digest keeps it current

A selected paper, video or article—linked to the lessons you need.

Practice makes it yours

Listen, check the idea, work on the language, then write for a new situation.

All connected to the same idea.

Here’s what learning actually looks like.

Try a small part of lesson one: does faster drafting mean a faster job?

A real lesson excerpt · F01
F01 · 16

The team to compare error rates, difficult cases, and customer outcomes before deciding whether the change is worthwhile.

Tap the phrase to see what it does.

Open this activity
WHY THE ORIGINAL

The same person. A more useful answer.

Meet Mina, a product manager planning a support assistant.

BEFORE · the first planning meeting
“We’ll train it on our docs so it knows the latest policy.”

The goal is clear. The mechanism and the check are missing.

AFTER PRACTICE · a new design review
“I’d lean toward retrieval: our docs change weekly. The trade-off is a slower reply. I’d want to rule out outdated sources.”
01
Recommend without overclaiming

I’d lean toward …

02
Make the trade-off explicit

The trade-off is …

03
Turn a risk into a check

I’d want to rule out …

What changed? She separates retrieving a document from changing model weights.

Lewis et al. · Retrieval-Augmented Generation ↗A worked example · not a learner testimonial
Try the RAG lesson
BEYOND THE LESSON

Leave with something you can use.

Three situations. Three complete examples of what to practise.

01 / A NOTE TO YOUR TEAM

Explain what matters.

A teammate shares an impressive AI demo. You add the question the team needs to ask.

EXAMPLE OUTPUT
The demo looks promising. Before we adopt it, I’d test it on our own tickets.
Practise evidence and evaluation
02 / A PILOT PROPOSAL

Turn an idea into a next step.

Your team wants a support agent. You define the first boundary.

EXAMPLE OUTPUT
Let’s start with drafting. A person approves each reply before it reaches the customer.
Practise agents and boundaries
03 / A DESIGN REVIEW

Make a recommendation you can defend.

A stakeholder asks why you chose RAG. You explain the trade-off and the test.

EXAMPLE OUTPUT
Our policy changes weekly. I’d use retrieval—and check for missing or outdated sources.
Practise retrieval and trade-offs
Explore the three situations

Illustrative scenarios · AI-generated artwork

THE SHIFT DIGEST

Worth your attention. Ready to learn.

A paper, a video, an engineering article. The newest reviewed guide in each format, with original dates shown below.

arXiv:2610.03195 2026
Source Preference in the Wild: How LLM Agents Favor Items by Source, and How to Reduce ItJonghyun Song, Haewon Park, Jeonghoon Shim, Woojung Song & Yohan Jo · Seoul National University
PAPER · ORIGINAL SOURCE
Paper01
Jonghyun Song, Haewon Park, Jeonghoon Shim, Woojung Song & Yohan Jo · Seoul National University

Would the choice change with a different source label?

When an AI assistant shortlists options for work, check whether its choice follows your requirements or a familiar name.

Intermediate · 12 min · suggested practice
Original · Reviewed
Read the guide
LightOnOCR-3: High-Performance OCR and Layout Extraction in One ModelLightOn · Said Taghadouini, Adrien Cavaillès and Baptiste Aubertin
Engineering03
LightOn · Said Taghadouini, Adrien Cavaillès and Baptiste Aubertin

Before summarizing a document, check what was read

A useful paper summary needs the right paragraph, table and units. Learn to spot extraction mistakes before they become confident answers.

Starter · 10 min · suggested practice
Original · Reviewed
Read the guide
All issues
THE ROADMAP

Choose what you want to be able to do.

Start with the shared foundations. Then take one route into your work.

Understand the technology

Explain a system choice

  1. 1Get your bearings
  2. 2Look inside the model
  3. 3Connect systems to evidence
See this learning route
Make product decisions

Propose a useful pilot

  1. 1Build a shared foundation
  2. 2Define and test the problem
  3. 3Decide how to adopt it
See this learning route
Improve your coding workflow

Ship a reviewed change

  1. 1Start with the foundations
  2. 2Read, change and test
  3. 3Ship with judgment
See this learning route
Already started? Find your place on the learning map.
BUILDERS TO FOLLOW

Follow the work. Then follow the builder.

A small reading list, chosen for what you can learn. Independent curation; no affiliation implied.

TALKS & ENGINEERINGAI Engineer ↗FOUNDERS & PRODUCTEO Studio ↗PRIMARY ENGINEERINGAnthropic Engineering ↗
Source guide
Explore the free lessons →