How to Use AI Tools for Studying: What's Allowed, What's Not, and What Actually Helps

The Shifting Landscape: Universities Are Embracing AI

The early panic is over. In 2023, many universities responded to ChatGPT with blanket bans. By 2025, most UK universities had moved to a more nuanced position:

• Russell Group statement (2023): AI literacy is a skill graduates need — universities should teach responsible use, not just prohibit it
• QAA guidance (2024): Assessment should be redesigned to account for AI, rather than relying solely on detection tools
• Most universities now use a "traffic light" system: Green (AI freely allowed), Amber (AI allowed with restrictions/declaration), Red (AI not permitted for this assessment)

The direction of travel is clear: AI is becoming a tool you're expected to use well, not something you're expected to avoid entirely.

Legitimate Use Cases: Where AI Helps You Learn

1. Brainstorming and Idea Generation

What to do: Ask an AI to help you generate initial ideas for an essay topic, explore different angles, or create mind maps.

Example prompt: "I need to write a 2,000-word essay on the impact of social media on democracy. Give me 5 possible angles I could take, with a brief description of each."

Why it's fine: You're not asking the AI to write your essay — you're using it to overcome the blank-page problem. You still choose the angle, do the research, and write the argument.

2. Concept Explanation

What to do: Ask AI to explain a concept you're struggling with in different ways until it clicks.

Example prompt: "Explain Foucault's concept of 'biopower' to me as if I'm a second-year sociology student. Then give me a real-world example."

Why it's fine: This is no different from watching a YouTube tutorial or asking a classmate to explain something. The AI is a learning aid, not a source you cite.

3. Practice Questions and Self-Testing

What to do: Generate practice questions based on your lecture notes or textbook content.

Example prompt: "Based on these key topics from my International Relations module — realism, liberalism, constructivism — generate 10 short-answer exam questions at second-year undergraduate level."

Why it's fine: Active recall is the most effective revision technique. AI lets you generate unlimited practice questions tailored to your exact syllabus.

4. Grammar and Language Checking

What to do: Use tools like Grammarly or AI to check grammar, punctuation, and sentence clarity in text you've already written yourself.

Why it's fine: This is the digital equivalent of having a native-speaker friend proofread your essay. You wrote the content; the tool polishes the language. Most universities explicitly allow this.

5. Coding and Technical Help

What to do: Use AI to debug code, explain error messages, or suggest approaches to programming problems.

Example prompt: "My Python function throws a TypeError when I pass a list. Here's the code: [paste]. What's wrong?"

Why it's fine: In professional software development, AI coding assistance is standard. Using it for learning and debugging is widely accepted, provided you understand the code you submit.

6. Research Navigation

What to do: Ask AI to suggest search terms, recommend seminal papers in a field, or explain the structure of a research area.

Example prompt: "What are the key debates in the literature on employee motivation? Suggest 5 seminal papers I should read."

Why it's fine: You're using AI as a starting point for your own research, not as a replacement for it. Always verify the papers actually exist (AI sometimes fabricates references) and read them yourself.

The Red Lines: What You Must NOT Do

Submitting AI-Generated Text as Your Own Work

This is academic misconduct at virtually every UK university. It doesn't matter whether you prompted the AI "well" or whether you edited the output. If the substance of the text was generated by AI and you present it as your own original work, it's a breach of academic integrity.

Using AI for Assessed Work Without Declaration

Even when AI use is permitted, many universities require you to declare how you used it. Failing to declare when required is itself a form of misconduct, even if the use was otherwise legitimate.

Using AI to Complete Exams

Unless explicitly permitted, using AI during supervised or timed assessments is cheating. This applies to in-person exams, take-home exams, and online tests.

Fabricating Sources

AI tools frequently generate citations that don't exist — they look real but the paper, journal, or author is made up. Submitting fabricated references is a serious academic offence. Always verify every reference your AI suggests against an actual database.

How to Declare AI Use

When your university requires an AI declaration, be honest and specific. Here's a template you can adapt:

I used [Tool Name, e.g., ChatGPT-4o] for the following purposes in this assignment: - Brainstorming initial essay angles (no AI-generated text was used in the final submission) - Grammar checking the final draft using Grammarly - Generating practice exam questions for revision (not included in the submission)

All ideas, arguments, analysis, and written content in this essay are my own original work.

For a more detailed guide on AI declaration statements, see our article on how to write an AI declaration.

Tool Comparison: Which AI Tools Are Worth Using?

ChatGPT (OpenAI)

Best for: General brainstorming, concept explanation, practice questions Strengths: Versatile, good at following complex instructions, widely available Limitations: Can fabricate sources; free version uses older model; doesn't cite real academic papers reliably Cost: Free tier available; Plus plan ~$20/month

Claude (Anthropic)

Best for: Long-form analysis, reading comprehension, careful reasoning Strengths: Handles long documents well, tends to be more cautious and accurate, good at nuanced analysis Limitations: Smaller ecosystem of plugins/integrations than ChatGPT Cost: Free tier available; Pro plan ~$20/month

Grammarly

Best for: Grammar checking, sentence clarity, tone adjustment Strengths: Real-time feedback in your browser, widely accepted by universities as a writing aid Limitations: Limited to surface-level language correction; premium features require subscription Cost: Free tier for basic grammar; Premium ~$12/month

Notion AI

Best for: Note organisation, summarising lecture notes, task management Strengths: Integrated into your existing notes and documents; good for structuring and organising Limitations: Not a standalone research tool; AI features are supplementary to the note-taking platform Cost: Free tier with limited AI; Plus plan ~$10/month

Perplexity AI

Best for: Research navigation, finding real sources with citations Strengths: Provides cited answers with links to real sources; better than ChatGPT for fact-checking Limitations: Less flexible for creative brainstorming; some features behind paywall Cost: Free tier available; Pro plan ~$20/month

Building AI Literacy as a Graduate Skill

Using AI well is increasingly listed as a desirable skill by employers. The key competencies:

1. Prompt engineering: Knowing how to ask the right question to get useful output
2. Output evaluation: Being able to critically assess whether AI output is accurate, complete, and appropriate
3. Ethical judgement: Understanding when AI use is appropriate and when it isn't
4. Tool selection: Choosing the right tool for the right task, rather than using one tool for everything
5. Integration, not replacement: Using AI to enhance your thinking, not to replace it

The students who thrive won't be those who avoid AI entirely, or those who rely on it blindly. It'll be those who use it as a thinking partner while maintaining their own critical judgement.

Common Mistakes When Using AI for Study

1. Treating AI Output as Fact

AI models generate plausible-sounding text, not verified truth. Always cross-check facts, statistics, and especially references against reliable sources.

2. Using AI as a Shortcut Instead of a Learning Tool

If you ask AI to summarise a chapter so you don't have to read it, you've saved time but learned nothing. Use AI to deepen your understanding, not to bypass it.

3. Not Understanding What You Submit

If you use AI to help with code or analysis, make sure you can explain every line. If your marker asks you to walk through your work and you can't, that's a problem regardless of how the work was produced.

4. Assuming All Universities Have the Same Rules

AI policies vary significantly between institutions, faculties, and even individual modules. Check your specific module's assessment brief and your university's AI policy before assuming something is allowed.

5. Forgetting That AI Detectors Exist

Tools like Turnitin's AI detection can flag AI-generated content. Even if you edit the output, stylometric analysis can detect patterns typical of AI writing. The safest approach: write it yourself and use AI only for the support tasks described above.

The Bottom Line

AI tools are powerful study aids when used correctly. Use them to brainstorm, explain concepts, generate practice questions, check grammar, and navigate research. Don't use them to generate the text you submit, fabricate sources, or complete assessments.

The golden rule: If removing the AI tool would remove your learning, you're using it wrong. If removing the AI tool would just make the process slower, you're using it right.

Stay informed about your university's specific AI policy, declare your use when required, and build AI literacy as a skill that will serve you long after graduation.