A ProximaED initiative — self-learning, self-assessment

AI Literacy
for all.

Five free, self-paced modules that take any engineering student — Mechanical, Civil, EEE, ECE, CSE, AI&ML, Robotics, Chemical, or otherwise — from "what is AI?" to using it well and checking its work. No account, no cost, no ML background assumed.

§

Independence notice. AI Literacy Lab is an independent educational initiative created and maintained by ProximaED. It is not affiliated with, endorsed by, or sponsored by Anthropic or any other AI company. Interpretive notes tied to specific interface behaviour (e.g. "why an assistant might use this word") are our own reading for learning purposes, not an official statement from any AI provider. For authoritative information about Claude, see anthropic.com and docs.claude.com.

01 — Five modules, one path

Self-learning, self-assessment, start to finish

Work through them in order, or jump to whichever you need. Each module ends where relevant with its own short quiz, and every quiz score is remembered on this device under Progress.

Module 1 · 34 Q&A

AI Fundamentals

What AI, ML, Deep Learning, GenAI, and LLMs actually are, how models are trained and evaluated, bias, hallucination, RAG, and how to apply AI responsibly as an engineer.

Start module →

Module 2 · 21 Q&A

How AI Works

The full pipeline from prompt to output — tokenization, embeddings, vectors, attention, and generation — with a step-by-step diagram and worked example.

Start module →

Module 3 · 151 entries

Interface Vocabulary

A field catalogue of the loading-screen words AI chat tools show you — cogitating, percolating, triangulating — with plain meanings and the metaphor each is borrowing from.

Start module →

Module 4 · 50 exercises

Prompt Lab

Hands-on engineering prompt-writing practice — mechanical and robotics scenarios where task clarity, context, and constraints matter more than clever wording.

Start module →

Module 5 · Interactive

Predict Game

Watch tokenization and next-token prediction actually run, built on real counted statistics from about a million tokens of public-domain text.

Start module →

Reference

Resources

Plain explanations of common AI-product features, a short FAQ, and links to academic and official sources for anything covered across these modules.

Open resources →

02 — An important correction

Does knowing this vocabulary make an AI reason better?

No — and this is worth being precise about, because it's the one place this whole lexicon could mislead you if left unsaid.

What these words actually are

Every word in the lexicon is interface status text — copy an AI chat product displays on screen while a response streams in. It's cosmetic. The underlying generation process is the same whether the screen says "Thinking…" or "Cogitating…". The word doesn't cause a different kind of reasoning; it's just what the loading state is labeled.

What if you use these words in your own prompt?

Different question, and here the honest answer is: it's a weak signal, not a switch. Asking a model to "cogitate deeply" can nudge its output — mainly by conditioning tone and framing — but it doesn't unlock a separate reasoning mode. The same request phrased plainly, with a clearer task and better constraints, will usually move the output further than a fancier verb will.

Weaker lever Vocabulary choice → mild semantic framing → slightly different tone
Stronger levers Clear task definition · sufficient context · explicit constraints · good examples · defined evaluation criteria · model choice · decoding settings → meaningfully different output

So a more accurate rewrite of the pattern this lexicon uses: not "cogitate makes Claude think deeply," but "cogitate is interface copy that reads as signaling deeper consideration — the actual quality of any response depends on the full prompt, task specification, context, and generation process, not the word choice." Every "why it might fit" note elsewhere on this page should be read with that in mind. If you're trying to actually improve output quality through better prompting, Anthropic's own prompt engineering guide is the real resource — it covers the levers that actually move outcomes.