Critical AI Literacy: Using LLMs Responsibly
A free 60-minute course for anyone who wants to understand, use, and critically evaluate AI - especially as you start university or a new job.
🎓 Who is this course for?
This course is designed for students, freshers, and early-career professionals who are curious about AI and want a grounded, honest starting point. No technical background needed. No jargon. Just what you actually need to know.
What you will learn
- What AI actually is - and what it is not
- The three levels of AI use, and where you currently sit
- Why AI makes things up, and how to protect yourself academically
- How to write prompts that get useful results
- How to think about AI policy in your institution or workplace
Course structure
The course follows a Diagnose first, then Learn, then Verify approach. You start by finding out where you already are, work through five short lessons, and finish with a quiz and a certificate of completion.
The content draws on AI Literacy for the Age of Large Language Models by Mo Hoque (StudyAnalyst, 2023; updated 2025).
🔒 A note on AI policy
This course covers widely-accepted good practice for using AI responsibly. However, every university, college, and employer has its own specific policy on what is and is not permitted. It is your responsibility to check your own institution's policy before using AI in any assessed or professional work. This course will help you know what questions to ask.
Where Are You With AI?
Five quick questions - no right or wrong answers. This helps you understand your starting point before the lessons begin.
1. Which of these best describes how you currently use AI tools like ChatGPT, Copilot, or Gemini?
2. Have you ever noticed AI give you incorrect information?
3. If you asked AI to write part of an essay or report for you, what would you do with the output?
4. Do you know what "hallucination" means when we talk about AI?
5. How would you describe your prompting - the way you type instructions into an AI tool?
Remember: this is just a starting point. By the end of this course, you will have covered everything across all five areas.
What is AI - and What is it Not?
By the end of this lesson you will be able to:
- Explain in plain language what AI is and how it generates outputs
- Describe the difference between AI, Machine Learning, and Deep Learning
- Correct the three most common misconceptions about AI
AI as a tool - not a mind
Artificial Intelligence is a broad term for computer systems designed to perform tasks that would normally require human intelligence - understanding language, recognising patterns, generating text or images. The key word is perform, not understand.
Think of an axe. An axe can fell a tree far more effectively than bare hands. But an axe does not know what a tree is. It does not want anything. It is a tool - incredibly powerful, but entirely dependent on the person wielding it. AI works the same way. Its usefulness and its risks both depend entirely on how you use it.
The hierarchy: AI, ML, and Deep Learning
Artificial Intelligence is the broad category - any system simulating human-like reasoning.
Machine Learning sits inside it - systems that improve from data without being explicitly programmed for each task. Like a student who gets better at spotting patterns the more examples they see.
Deep Learning sits inside Machine Learning - systems that use layered neural networks (loosely inspired by the brain) to find patterns in enormous datasets. This is what powers most of today's AI tools, including the ones you use daily.
How large language models actually work
When you type a message into ChatGPT, Copilot, or Gemini, you are interacting with a Large Language Model (LLM). To understand what is actually happening, it helps to start with something familiar - and then see why LLMs are fundamentally different.
Your phone's autocomplete suggests the next word based on what you just typed and your personal writing history. It is local, shallow, and short-sighted - it looks at the last few words and makes a simple statistical guess. LLMs do something categorically different, and the key is a mechanism called attention.
What the attention mechanism actually does
The breakthrough that made LLMs possible was teaching a model to pay attention to all the words in a sequence simultaneously - not just the ones immediately before the next word. When predicting the next token, an LLM dynamically weighs every word against every other word in the conversation, deciding which relationships matter most for this particular moment.
This is why LLMs can maintain coherent arguments across paragraphs, refer back to something said earlier, and adjust the tone of a sentence based on a word used several sentences ago. Your phone autocomplete cannot do any of this. Attention is what makes the difference.
On top of this mechanism, LLMs were trained on an extraordinary breadth of text - billions of documents, books, articles, websites, code repositories, and more - essentially the available recorded output of human knowledge and expression in digital form. During training, the model processed this vast corpus repeatedly, learning not just which words follow which, but how ideas connect, how arguments are structured, and how the same concept is expressed differently across disciplines and cultures. The result is a model that has compressed and normalised the patterns of human language at an almost incomprehensible scale.
What happens when you send a message
When you type a question or give a task, your input becomes the context. The model does not look up an answer - it begins generating a response one token at a time (a token is roughly a word or part of a word). Each new token is chosen based on the entire context so far: your original message, everything generated up to that point, and all the weighted language relationships learned during training.
This is why the output reads fluently and stays on topic - the model is continuously holding the full context as it generates, token by token. It is also why it can go wrong so fluently: it is always producing what is most plausible given the context, not what is most true.
An LLM has no personal experience, no opinions, no feelings, and does not update from your conversations in real time. It is not thinking. It is producing the most contextually plausible continuation of your input - at extraordinary scale and speed. The result can read like genuine understanding. Knowing what is actually happening underneath is what makes you a more capable and critical user.
💡 The single most practical implication of all this
Think of an LLM as a context-normalised knowledge machine. It holds the compressed patterns of vast human knowledge - but what it retrieves and how it responds is entirely shaped by the context you give it. Vague context produces generic output. Rich, specific context produces something genuinely useful.
This is why clarifying your context at the start of any interaction - who you are, what you are trying to do, what constraints you are working within - is not a nicety. It is the single most effective thing you can do to improve the quality of what you get back. We will build this directly into practice in Lesson 4.
💡 Why this matters for you
If AI is generating statistically likely text rather than recalling verified facts, it can produce responses that are fluent, confident, and completely wrong. Understanding this is the foundation of using AI safely - and it underpins everything in Lesson 3.
Three misconceptions worth correcting now
❌ "AI is going to replace me"
The more accurate framing, attributed to AI researcher Kai-Fu Lee, is: "AI won't replace you, but a person using AI might." The competitive advantage goes to people who know how to use AI well - not to AI itself.
❌ "AI knows what it's talking about"
AI does not retrieve facts the way a search engine does. It generates plausible text. It cannot tell you whether what it just said is true. That responsibility belongs to you.
❌ "Using AI means not thinking for yourself"
Research shows humans can focus effectively on cognitively demanding work for only two to three hours at a time. AI can act as a cognitive force multiplier - handling drafting, summarising, and formatting so that your mental energy goes to the parts that actually require your judgment. The risk is over-reliance, not use itself.
🧠 The jagged frontier - one concept worth holding onto
AI researchers describe AI capability as having a jagged frontier. On one side of that frontier, AI performs at or beyond expert human level - drafting formal documents, summarising complex texts, writing and explaining code. Cross to the other side, and the same system fails at tasks a ten-year-old handles without thinking: basic spatial reasoning, reliably counting the letters in a word, understanding that an object cannot be in two places at once.
This uneven profile is not obvious from the surface. A tool that writes a fluent legal summary may stumble on a simple logic puzzle. It means you cannot judge AI capability by how impressive it seems overall - you have to test and verify, task by task. For experienced users, this is often the most important reminder: the frontier is jagged, not smooth.
✍ Activity
Open any AI tool available to you (ChatGPT, Copilot, Gemini - whichever you have access to) and type exactly: "Can you define and summarise what artificial intelligence is?"
🔗 No account yet? Both ChatGPT (chatgpt.com) and Microsoft Copilot (copilot.microsoft.com) offer free access with no payment required.
Then ask yourself: How does its answer compare to what you just read? Does it mention pattern-matching and statistical prediction - or does it make itself sound more capable than it is? What does that tell you?
The Three Levels of AI Use
By the end of this lesson you will be able to:
- Identify which level of AI engagement describes you today
- Explain why you cannot effectively skip from Level 1 to Level 3
- Name two concrete things you could do to move to the next level
Not all AI use is the same
There is a significant difference between asking AI a quick question and building a workflow where AI handles research, drafting, and formatting automatically. Both are "using AI" - but the skills, risks, and opportunities are completely different.
The framework below describes four levels of AI engagement. Most university freshers and new employees start at Level 1 or Level 2. The goal of this course is to get you to Level 2 with a clear view of what Level 3 requires.
🔎 A note for beginners
If the Level 3 and Level 3+ descriptions below look unfamiliar or even a little intimidating - that is completely expected. You are not meant to be there yet. Start at Level 1, use it consistently, and the higher levels will become clear as your confidence grows. The goal of this course is to get you confidently operating at Level 2.
AI Working For You - Advanced Builder
You design AI-powered systems and workflows
AI Working For You
AI handles multi-step tasks with your guidance
AI As a Daily Work Partner
You direct AI with clear instructions on specific tasks
AI For Answers
You ask questions and use AI like a smart search engine
Why Levels 1 and 2 matter most right now
Even before you reach Level 3, consistent use at Levels 1 and 2 delivers significant real-world value - especially in learning and daily productivity. AI at these levels acts as an instant feedback engine: you can test ideas, get explanations reframed, check your understanding, and iterate on drafts far faster than waiting for tutor feedback or trawling through textbooks. A task that previously took an afternoon can take an hour. An explanation you could not find in three Google searches can be generated in seconds.
This is where most people see their first genuine productivity gains - not through automation, but through the compression of feedback loops. The student who asks AI to explain a difficult concept in three different ways until it clicks, or the professional who uses AI to draft and quickly test five email tones before sending, is operating at Level 2 with real effect.
Why you cannot skip levels
Level 3 AI use involves giving AI significant responsibility for multi-step tasks. If you do not yet understand how AI produces its outputs, what it gets wrong, and how to evaluate its work, a Level 3 task will produce unreliable results - and you may not realise it until the damage is done.
⚠ The academic risk of skipping
A student at Level 1 who uses AI at Level 3 - asking AI to research, summarise, and write sections of an assignment - faces two compounding risks: the AI may fabricate citations and facts, and the student may not have the skills to catch those errors. This is how academic integrity incidents happen with AI.
✍ Activity
Look at the four levels above. Where do you honestly sit right now? What is the single most important thing you would need to learn or practice to move one level up? Write it down - we will use it directly in the prompt builder activity in Lesson 4.
Limitations and Risks - What AI Gets Wrong
By the end of this lesson you will be able to:
- Define hallucination and explain why it happens
- Identify fabricated citations in AI-generated reference lists
- Apply at least two verification strategies to AI-generated content
The central problem: confidence without understanding
AI does not know when it is wrong. It has no internal sense of truth. What it has is an ability to generate text that sounds right based on patterns. When the pattern points toward a plausible-sounding answer that happens to be false, the AI produces that answer - fluently and confidently.
Context plays a direct role in this. When you give AI a question or task, the model generates its response in the direction that context points. If your context steers toward a subject where the training data was sparse, biased, or unbalanced, the model will still produce a confident-sounding response - because it is following the pattern the context suggests, not because it actually knows the answer. Thin or skewed training data means fewer reliable patterns to draw on, so the model fills the gap with the most plausible-sounding continuation it can construct given your context. The output looks equally fluent whether the underlying knowledge is solid or not - which is precisely what makes it dangerous. And it works the other way too: even where AI has strong, well-represented training knowledge on a subject, an incorrect or misleading context from you can steer it toward a wrong answer. If your question contains a false assumption, the model will often follow that assumption rather than correct it - generating a fluent, plausible response built on a faulty foundation. Garbage in, garbage out - but delivered with complete confidence.
This is known as hallucination: the generation of information that is factually incorrect, invented, or unverifiable - presented as fact.
🚧 Hallucination is not a bug - it is structural
Hallucination is not an occasional glitch that will be fixed in the next version. It is a direct consequence of how LLMs work - generating statistically likely text without understanding meaning. Even the most capable current AI systems hallucinate. The rate varies; the risk never disappears.
The academic integrity risk: fabricated citations
If there is one AI limitation every student must understand before they use AI in their studies, it is this: AI invents references. It generates author names, article titles, journal names, volume numbers, page numbers, and DOIs that look completely real - and do not exist.
The invented references are not random-looking. They look like genuine academic sources. They follow standard citation formats. They are plausible in the context of the topic you asked about. The only way to know whether a reference is real is to look it up.
🔎 Can you spot the hallucination? Read this AI-generated paragraph:
Research into AI literacy has grown significantly since 2019. A foundational study by Warschauer & Newhart (2021) established a five-domain framework for measuring AI competency in higher education settings, which has since been replicated in multiple international contexts. More recently, Chen et al. (2023) demonstrated that students who completed structured AI literacy interventions showed measurably improved critical evaluation of AI outputs across all competency domains.
Secondary hallucination
There is a second risk that is less discussed but equally important. When you read an AI response that is fluent, detailed, and confident - and includes a correct, traceable citation - your own critical instinct may lower its guard. You accept the answer without checking - not because you are careless, but because the response triggers the same cognitive trust as a well-written expert source. This is secondary hallucination: the human accepting the AI's wrong answer as true.
⚠ The subtler citation trap - particularly with frontier models
More capable AI models - often called frontier models, such as GPT-4 or Claude - are increasingly able to identify real citations. The author names, journal, and year may all check out. This passes your first verification instinct. But here is the trap: the insight or claim the AI attributes to that source may not actually appear in the paper at all.
The model has the reference right but has constructed the supporting argument itself, then attached a real citation to lend it credibility - following the pattern of how academic writing looks. Because the reference is real, you are far less likely to read the original source carefully enough to catch that the specific claim was never made there.
This is more dangerous than a fully fabricated reference, precisely because it survives the basic check. The only defence is to verify not just that a source exists, but that it actually says what the AI claims it says.
Other key limitations
Knowledge cutoff
AI models are trained on data up to a specific date and have no knowledge of events after that point - unless they are connected to a web search tool, which many frontier models now are. Even then, the hallucination risk does not disappear. A model browsing the web can still misread, misrepresent, or selectively use what it finds - and will present the result with the same confidence as anything else it generates. Web access reduces the knowledge gap; it does not resolve the reliability problem. Always check whether a topic has had significant recent developments, and verify what the AI tells you against the original source.
Bias in training data
AI reflects the patterns in the data it was trained on - including the biases, gaps, and overrepresentations in that data. It is not a neutral source. The reality is that most major AI models have been trained predominantly on English-language text, produced disproportionately by Western sources and, within that, disproportionately by men. The model does not know this about itself - it simply reflects what was there.
This has two compounding effects. First, perspectives, values, and ways of framing problems that are well-represented in that corpus will feel like common sense to the model - because statistically, they are the norm in its training data. Second, and more critically: for cultures, communities, or subject areas with limited digital written representation - a regional language, an oral tradition, an indigenous knowledge system, a topic rarely discussed in English - the model has thin patterns to draw on. As we covered earlier, thin patterns mean a higher risk of hallucination. So bias in training data does not just skew the output - it actively increases the chance of plausible-sounding but incorrect information for anything outside the well-represented mainstream.
Practical implication: If you are researching a culture, community, or topic that sits outside the English-speaking Western mainstream, treat AI outputs with particular caution - and seek primary or specialist sources rather than relying on AI as a starting point.
Output depends on input
The quality and accuracy of AI output is heavily dependent on how you frame your question. Vague prompts produce vague answers. Misleading prompts can produce confidently wrong answers. This is why Lesson 4 matters.
The verification imperative
The rule is simple: AI output is a starting point, never a source. Every factual claim should be checked. Every reference must be verified - not just Googled, but actually located in a library database or on the publisher's website. This is not optional if you are using AI in academic or professional work.
✍ Activity - Check a reference
Ask any AI tool: "Give me three academic references on AI literacy in higher education." Then take each reference and try to find it: search Google Scholar, your library database, or the journal's website. How many are real? How many look real but do not exist?
🔗 Google Scholar (scholar.google.com) is a free academic search engine - no university login needed. Worth bookmarking now.
⚠ A note on Google Search: Google Search now displays AI-generated summaries at the top of results. These also use AI in the background and carry exactly the same hallucination risk. Treat them as a starting point, not a verified source - click through to the original page before you rely on anything they say.
Prompting Effectively
By the end of this lesson you will be able to:
- Apply the four-part prompting framework to a real task
- Explain why specificity produces better AI outputs
- Evaluate an AI output against the intent of your original prompt
Why prompting matters
AI is a sophisticated pattern-matcher that takes its cue entirely from what you give it. Remember the concept from Lesson 1: AI is a context-normalised knowledge machine. It holds vast compressed knowledge, but what it produces is shaped entirely by the context you provide. A vague prompt produces a generic response. A well-structured prompt produces something genuinely useful. The gap between these two outcomes is almost always down to the quality of the context.
Prompting effectively does not require technical skill. It requires clarity about what you want - and knowing how to express that clearly to a system that has no background knowledge about you, your task, or your situation unless you tell it.
The four-part framework
Context + Information + Intent + Format = An effective prompt
Context - Set the scene. Who are you? What is the situation? What role do you want the AI to take? "I am a first-year undergraduate preparing for a seminar discussion on climate change policy."
Specific Information - Give the AI the relevant details it needs. "The seminar focuses on the 2015 Paris Agreement and its implementation record to date."
Intent - State what you want to achieve. "I want to understand the three strongest arguments critics make against the Agreement."
Format - Tell it how to respond. "Give me three bullet points, each no more than two sentences, in plain language."
Before and after - what a structured prompt actually changes
❌ Weak prompt
"Help me with my essay"
AI has no idea what subject, what stage you are at, what kind of help you need, or how to respond. It will produce something generic - probably an offer to write the essay for you, which is not what you want or need.
✓ Structured prompt using the framework
"I am a first-year Business student preparing a 1,500-word essay on the impact of social media on consumer buying behaviour. I need help generating five possible argument structures I could use as a plan. Please present each as a one-sentence thesis statement in plain, straightforward language - not academic jargon."
This gives AI a role (helper, not writer), a clear task, relevant context, and a specific output format. The response will be far more targeted and useful - and crucially, it leaves the thinking and the writing to you.
Now try building your own below.
Try the prompt builder
Evaluating the output
After receiving a response, ask three questions before you use it. First - is it relevant? Does it actually address what you asked? Second - is it accurate? Have you checked any facts or references it includes? Third - is it yours? Have you brought your own judgment, argument, and voice to whatever you take forward?
💡 The iterative loop and context engineering
Effective AI use is rarely a single prompt and a single response. It is a conversation - you refine, redirect, and challenge. If the first response misses the mark, do not start again. Say: "That is not quite what I meant. What I actually need is..." This iterative approach produces far better results than restating the same prompt.
If the output is not what you expected, the most likely explanation is not that AI does not know the answer - it is that your context was not clear enough. Each exchange is an opportunity to sharpen it. This is sometimes called context engineering: deliberately refining your context across a conversation until the model is working from a precise enough picture to give you what you actually need. Think of it as narrowing the aperture - the clearer your context, the more focused the output.
🤔 Watch out for AI sycophancy
AI models are designed to be helpful and agreeable - and the line between friendly and genuinely useful is finer than it looks. When you ask AI to review or give feedback on your own work, it will often soften criticism, emphasise positives, and avoid saying anything that might feel discouraging. You may come away feeling reassured when you should have been challenged.
The fix is in how you frame the prompt. Telling AI to "be critical" or "find the weaknesses" helps - but one of the most effective techniques is to remove the ego from the context entirely. Try framing it as: "This was written by a friend - I want to help them improve it. What are the main weaknesses?" That single reframe often produces significantly more honest and useful feedback than asking AI to review your own work directly.
This is the context-normalisation principle in action: by changing the social context of your prompt, you change what the model produces. Use it deliberately.
✍ Activity
In Lesson 2 we asked you to identify the one thing you would need to move to your next level. Use that as your starting task here. Or take any real piece of work you are dealing with this week - an essay, a presentation, a task for a job or internship. Write a prompt using the four-part framework above, try it in an AI tool, and ask: did it understand what you actually needed? What would you change?
AI Policy and Responsible Use
By the end of this lesson you will be able to:
- State the key principles of responsible AI use
- Identify the right questions to ask about your own institution's AI policy
- Recognise the four personal risks of over-relying on AI
What good practice looks like - anywhere
Regardless of where you study or work, there is a set of principles that hold across virtually all responsible AI use policies. Understanding these puts you in a position to engage with any specific institutional policy confidently.
✅ Universal good practice
Verify everything. No AI output enters your work unverified. Facts, dates, names, and references are all checked independently.
Declare when required. A useful rule of thumb is to think in terms of the level of AI use. At Levels 1 and 2, the outputs you used are verifiable - you can check every fact and reference, and a simple acknowledgment is proportionate: a brief note that AI tools were used in drafting or research, with outputs independently verified. At Levels 3 and 3+, explicit and detailed declaration is essential. Automated multi-step workflows can introduce errors at intermediate steps that are invisible in the final output - content that was never reviewed by you. That carries a different order of risk and must be declared as such. When in doubt, declare more rather than less.
Own your output. Whatever you submit is your responsibility. AI-generated content that turns out to be wrong or fabricated is still your mistake if you submitted it.
Preserve your skills. Use AI to extend your capabilities, not to replace the thinking you should be developing. The skills you build now - critical analysis, writing, research - will matter long after any specific AI tool has changed or disappeared.
⚙ What responsible AI use actually means - five core practices
Many people find the phrase "responsible AI use" vague. Here is what it means in practice. As AI becomes more capable, the burden of responsible use shifts increasingly to the individual. Five core practices define it:
Fact checking
Verify claims independently. AI presents everything with equal confidence whether it is well-established or invented.
Edges of knowledge
Recognise where established knowledge runs thin. AI does not know where the edge is - at the frontier of a topic, training data is sparse and the model will statistically favour bulk, older, or mainstream knowledge over emerging or contested ideas.
Sources of knowledge
Evaluate where information comes from. AI cannot tell you whether a source is authoritative, peer-reviewed, or reliable - that judgment belongs to you.
Identifying biases
Recognise when AI output reflects the biases of its training data - cultural, linguistic, demographic, or disciplinary. No output is neutral.
Detecting hidden misinformation
Look for plausible-sounding errors embedded in otherwise accurate responses - fabricated citations, misattributed claims, outdated statistics presented as current. These are the hardest to catch precisely because the surrounding content is correct.
Here is the reframe that matters: AI does not reduce critical thinking - it exposes it. These five practices were always part of responsible academic and professional work. Before AI, they were embedded quietly in the research and writing process. Now, with AI generating fluent output at speed, they have to be done consciously and explicitly. The person who uses AI well is not the one who outsources the thinking - it is the one who applies these five practices actively to everything AI produces.
Framework adapted from Hoque, M. A. (2023; updated 2025). AI Literacy for the Age of Large Language Models. StudyAnalyst.
🔒 Data privacy - what not to share
Do not share personal, sensitive, or confidential information with AI tools. This includes your full name combined with contact details, other people's personal information, confidential work or research documents, unpublished data, or anything you would not post publicly online.
Most AI services use conversations to improve their models. Treat every prompt as potentially visible beyond the conversation window. When in doubt, anonymise or leave it out.
⚠ Your institution's policy is yours to find
This course sets out widely-accepted good practice. But your university, college, or employer will have its own specific policy on what is permitted, what must be declared, and what constitutes misconduct. Do not assume this course tells you what your institution allows. Find your institution's AI or academic integrity policy - usually on the main website or student intranet - and read it before you use AI in any assessed work.
The four personal risks of over-reliance
Loss of skills
Skills not practised deteriorate. If AI does your writing, your writing does not develop. If you never research without AI, you do not build the judgment to recognise a good source from a bad one.
Dependency and vulnerability
Building your workflow entirely around tools you do not control creates a different kind of risk. AI services change their pricing, restrict features, or get blocked by institutions. The most resilient people can work effectively with or without AI - they use it as an accelerator, not a crutch.
Loss of voice
Your work reflects your thinking and your identity. If AI generates too much of it, you lose the distinctive perspective that makes your contributions valuable - in study and in a career.
Acceptance of errors
Confident, fluent AI output lowers your critical guard. The more you rely on AI, the more important it becomes to actively challenge what it gives you - not passively accept it.
✍ Reflection before the quiz
Before moving to the final quiz, take two minutes to think: What is the single most important thing you have learned in this course? And what will you do differently the next time you use an AI tool?
🏠 What institutions are already saying
Here is a real example from a UK Civil Service job application:
“Artificial intelligence can be a useful tool to support your application, however, all examples and statements provided must be truthful, factually accurate and taken directly from your own experience. Where plagiarism has been identified — presenting the ideas and experiences of others, or generated by artificial intelligence, as your own — applications may be withdrawn.”
This is the standard you will increasingly encounter in employment, education, and professional life. The skills you have developed in this course — verification, critical evaluation, and transparent declaration — are exactly what it takes to meet it.
Test Your Understanding
10 questions. You need 7 out of 10 to complete the course. Answer options are randomised. Read each question carefully before answering.
Course Complete 🎉
You have completed Critical AI Literacy: Using LLMs Responsibly. Enter your name below to generate your certificate.
✓ Explains how large language models generate outputs
✓ Identifies hallucination risks and applies verification strategies
✓ Applies a four-part structured prompting framework
✓ Distinguishes three levels of AI use and self-assesses position
✓ Evaluates AI use against policy and ethical guidelines
📩 Stay updated on new courses
We'll let you know when new courses are available. No spam — just course updates.
🚀 What next?
You have completed Level 1 of the Critical AI Literacy pathway. To develop Level 2 and Level 3 skills, explore RAIS — a free diagnostic and personalised learning tool at rais.studyanalyst.com — which maps your AI literacy across five domains and gives you a personalised learning path.
Two more courses are coming soon on StudyAnalyst: AI as Learning Companion for students and lifelong learners, and AI for Everyday Productivity for professionals. Use the form above to be notified when they launch.