AI Literacy Test

AI Literacy Test

Nearly everyone uses AI now. Far fewer people could explain what it is actually doing, which is exactly how you end up trusting a confident answer that happens to be wrong.

This free AI literacy test asks 20 questions about how artificial intelligence really works. Prompts, hallucinations, training data, bias, deepfakes, privacy, and the gap between what today's tools can do and what people assume they can do.

Every question explains itself the moment you answer, so you learn as you go rather than waiting until the end. You get a score out of 100, a level, and a free certificate with your name on it. No sign up, no email, and nothing you answer leaves your browser.

How the AI Literacy Test Is Scored

Twenty questions, each worth five points. Your score is simply the share you got right, and the answer choices reshuffle every time you take it.

Correct answersScoreLevel
18 to 2090 to 100AI Expert
15 to 1775 to 85AI Savvy
12 to 1460 to 70AI Capable
8 to 1140 to 55AI Aware
7 or fewerBelow 40AI Curious

Every question explains itself as soon as you answer, so a low score on the first run is not really a failure. Most people finish knowing several things they did not know when they started.

The AI Terms Worth Understanding

These are the ideas the test covers. If any of them were new to you, this section is the short version.

How the tools actually work

TermWhat it means
Artificial intelligenceSoftware that handles tasks which normally need human thinking, such as recognizing speech, sorting images, or writing text.
Generative AIAI that produces new content rather than retrieving existing content. This is the category chatbots and image tools belong to.
Large language modelThe engine behind text chatbots. It predicts likely text one piece at a time based on patterns learned from an enormous amount of writing.
Training dataThe material a model learned from. Its strengths, blind spots, and biases all trace back here.
Training cutoffThe point where a model's learning stopped. Anything after that is unknown to it unless the tool can search the web.
Narrow AIWhat exists today. Systems built for particular kinds of tasks, without self awareness or general understanding.
AGIArtificial general intelligence. A hypothetical system able to match human ability across most tasks. It does not exist yet.

The prediction point is the one most worth absorbing. A chatbot is not looking up an answer in a database and reporting back. It is generating text that fits the patterns of what usually follows your question. Most of the time the most plausible continuation is also the true one, which is why it works so well. The trouble starts when it is not.

Where AI goes wrong

TermWhat it means
HallucinationWhen a model states something false with complete confidence. Invented citations, wrong dates, and plausible sounding names are common examples.
BiasSkewed output caused by patterns in the training material. The model is not holding opinions, it is reflecting what it absorbed.
DeepfakeSynthetic images, audio, or video made to look like a real person said or did something they did not.

Hallucination is the single most important concept on this page, because it is invisible. A wrong answer arrives in the same calm, well organized tone as a right one. There is no wobble in the voice and no hedging unless the model was built to hedge. That is why the habit of checking anything that matters is worth more than any clever prompt trick.

Using AI safely

IdeaWhy it matters
Your dataDepending on the tool and its settings, what you type may be retained or reviewed. Treat a public chatbot like a stranger, not a private notebook.
Your responsibilityIf you use AI output at work or in public, you own it. The tool is not accountable for errors, you are.
PromptThe instruction you give. Better input reliably produces better output.

What AI Is Good At, and What It Is Not

Knowing where the line sits saves a lot of wasted effort and a fair amount of embarrassment.

Works wellHandle with care
Drafting a first version of somethingAnything where an error carries real cost
Summarizing long material you can checkSpecific facts, figures, dates, and quotes
Brainstorming and generating optionsCitations and sources, which are frequently invented
Rewriting, reformatting, changing toneMedical, legal, and financial decisions
Explaining concepts in simpler languageVery recent events, unless the tool can search
Code snippets you are able to testAnything involving your private or confidential data

The pattern is fairly simple. AI is strong when you can judge the output yourself and weak when you are relying on it precisely because you cannot. That is uncomfortable, because the second case is exactly when it feels most useful.

How to Write Better Prompts

Most disappointing AI results come from thin instructions rather than a weak model. Four habits fix the majority of it.

  • Say who it is for and why. A summary for a colleague and a summary for a customer are different jobs.
  • Give the shape you want. Length, format, tone, whether you want bullet points or flowing paragraphs.
  • Provide the source material. Pasting in the actual text beats asking the model to recall something it half remembers.
  • Ask for revisions. The first draft is a starting point. Saying what is wrong with it works better than starting over.

One extra habit is worth building: ask the model how confident it is and what it is unsure about. It will not always be right about its own uncertainty, but it often flags the parts most worth checking.

Spotting AI Generated Content

This is harder than most advice suggests, and it is getting harder.

Automated AI detectors are unreliable. They produce false positives on genuine human writing often enough that treating their output as proof is a mistake, and that mistake has already caused real problems for students and writers accused on the strength of a detector score.

For images and video, the old tells like strange hands and garbled background text are disappearing fast. The more durable approach is context rather than pixels. Where did this come from? Is anyone credible reporting the same thing? Does the account that posted it have a history? A surprising clip of a public figure saying something explosive deserves a search before a share, every time.

The field moves quickly, so specific tools, model names, and capabilities change faster than any article can keep up with. The underlying ideas on this page, prediction rather than lookup, hallucination, training cutoffs, bias, and your own responsibility for output, have held steady and are likely to keep holding.

Frequently Asked Questions

What is AI literacy?

Understanding enough about how AI works to use it well and judge its output sensibly. It does not require technical skill. It mostly means knowing what these tools do, where they fail, and when to check.

What is an AI hallucination?

When an AI presents false information as fact, confidently and fluently. It happens because the system produces text that fits a pattern rather than looking up a verified answer. Made up sources and citations are among the most common examples.

What is a prompt?

The instruction or question you type into an AI tool. Clear, specific prompts with context produce noticeably better results than vague one line requests.

What is the difference between narrow AI and AGI?

Narrow AI is what we have. It handles particular kinds of tasks and has no general understanding or awareness. AGI would match human ability across most tasks. It remains hypothetical.

Why does AI not know about recent events?

Models are trained up to a cutoff date and have no knowledge of what came after. Some tools can search the web to fill that gap, which is why one assistant may know about something another does not.

Is it safe to put personal information into AI chatbots?

Be cautious. Depending on the service and your settings, conversations may be stored or reviewed. Keep passwords, financial details, medical records, and anything confidential out of public tools, and check the provider's terms and privacy settings for anything sensitive.

Can AI replace my job?

It is changing many jobs more than it is erasing them, particularly roles heavy on routine writing, summarizing, and first drafts. The more useful question is which parts of your work AI handles poorly, since that is where your value concentrates.

Is this test free, and do you store my answers?

Free, with no sign up or email. Everything runs in your browser and nothing you select is saved or transmitted.

More Free Tests

  • AI Literacy TestYou are here. Twenty questions on how AI works, with instant explanations and a free certificate.
  • Online Safety TestCheck your instincts on scams, passwords, phishing, and privacy.
  • Critical Thinking ExerciseTwelve real world scenarios that train you to spot weak reasoning.
  • Data Analytics TestHow well you read numbers, charts, and the claims people build on them.

Free AI and Tech Tools