AI & Academic Integrity
AI & Academic Integrity
Verify claims, data and references using authoritative sources.
Look for fabricated information, omissions and bias.
Consult and cite original sources rather than AI summaries.
AI can produce useful ideas and explanations, but it can also generate information that is inaccurate, incomplete, outdated or biased. Its language may sound confident and authoritative even when the content is wrong.
An AI response is not evidence. You are responsible for verifying any AI-generated information that informs your work.
DP work requires you to select, evaluate and interpret evidence independently. AI may help you identify questions or possible directions when permitted, but it cannot replace engagement with subject-specific sources, data, texts or scholarship.
You must be able to explain:
Why you selected each source.
How you determined that it was reliable and relevant.
What evidence supports your claims.
Where interpretations or sources disagree.
How you reached your own conclusions.
Break an AI response into individual claims before evaluating it.
Check:
Names, dates, places and events
Quotations and attributed ideas
Statistics, calculations and datasets
Scientific explanations and terminology
Historical interpretations
Summaries of literary or artistic works
Legal, political or economic information
References to books, articles, websites or studies
Claims about current events or recent developments
The more important a claim is to your argument, the stronger the supporting evidence should be.
Verify important information using sources appropriate to the subject, such as:
Books and library databases
Peer-reviewed academic journals
Government and intergovernmental organisations
Universities, museums and recognised research institutions
Established news organisations
Primary and original sources
Published datasets and official statistics
Course texts and teacher-approved materials
Use the source itself—not the AI’s description of it—to determine what it actually says.
AI tools may provide links or references, but these are not automatically reliable.
When an AI tool mentions a source:
Locate and open the original source.
Confirm that the author, title and publication details are accurate.
Read enough of the source to understand its context.
Check that it genuinely supports the claim.
Evaluate the source’s authority, evidence, purpose and currency.
Cite the original source directly.
Do not cite a source that you have not opened and examined.
An AI hallucination occurs when an AI tool generates information that appears believable but is inaccurate or invented.
AI may create:
Events that never occurred
Quotations that no one said
Statistics without supporting data
Experiments or findings that were never published
Books, articles, authors or websites that do not exist
References containing a mixture of real and invented details
Incorrect explanations of genuine sources
If you cannot locate or verify a claim, do not use it as evidence.
A statement can be factually correct but still provide weak support for a conclusion.
Ask:
What evidence is provided?
Is the evidence relevant to the claim?
Is the evidence sufficient?
Is it current enough for the purpose?
Is the source qualified to address the subject?
Has the evidence been taken out of context?
Does the response confuse correlation with causation?
Are limitations or uncertainties acknowledged?
Do other credible sources reach different conclusions?
Do not accept a conclusion simply because some of the surrounding facts are correct.
An AI response can mislead by leaving out important context.
Consider:
Has the response oversimplified a complex issue?
Are important causes, consequences or limitations missing?
Does it present a contested interpretation as settled fact?
Does it confuse fact, interpretation and opinion?
Has it ignored evidence that challenges its conclusion?
Are the examples representative or selectively chosen?
Has the information become outdated?
Does the response use terminology accurately?
Investigate what is absent as carefully as what is included.
AI systems learn from existing data and publications. These materials may contain stereotypes, inequalities, dominant narratives and biased assumptions.
Ask:
Does the response favour one position without justification?
Does it present one cultural viewpoint as universal?
Does it make assumptions about a person or group?
Does it reproduce stereotypes or generalisations?
Does its wording influence how the reader views the issue?
Are commercial, political or institutional interests involved?
Does it treat an opinion or value judgement as a fact?
Bias is not always obvious. Compare the response with sources produced from different positions and contexts.
One AI response cannot represent every experience, discipline or interpretation.
Consider:
Whose voices are included?
Whose voices are absent or marginalised?
Are local, cultural and multilingual perspectives represented?
Are people directly affected by the issue heard?
Are Indigenous or historically underrepresented perspectives considered?
How might the issue be understood in another country or community?
Do credible scholars or sources disagree?
Seek evidence that broadens, complicates or challenges the initial response.
Different DP subjects require different forms of evidence.
Sciences
Verify scientific claims through peer-reviewed research.
Examine methodology, sample size, variables and limitations.
Do not use invented data, results or references.
Individuals and societies
Distinguish evidence from interpretation.
Compare primary and secondary sources.
Consider historical context, perspective and competing explanations.
Language and literature
Return to the complete original text.
Check quotations, context and translation.
Develop your own interpretation rather than adopting an AI reading.
Mathematics
Check every stage of the reasoning and calculation.
Test whether assumptions and methods are valid.
Do not accept a correct-looking answer without understanding the process.
The arts
Verify information about artists, works, movements and contexts.
Distinguish description from interpretation.
Consider authorship, copyright and cultural context.
TOK
Identify assumptions, claims and counterclaims.
Examine how the response constructs or justifies knowledge.
Investigate whose knowledge and perspectives are privileged.
Use this process whenever you evaluate an AI response:
Identify the claims that require checking.
Locate reliable, relevant and subject-appropriate sources.
Read and evaluate the original sources.
Compare the AI response with the available evidence.
Investigate errors, bias, disagreement and missing perspectives.
Revise your understanding using what you have verified.
Develop your own interpretation or conclusion.
Cite the original evidence appropriately.
If you quote, adapt or include AI-generated material, identify and reference that material as required. A citation to an AI tool does not replace citations to the original sources supporting factual claims.
Before using information from an AI response, check:
Can I confirm this information independently?
Are the supporting sources genuine and authoritative?
Have I opened and examined each source?
Does each source genuinely support the claim?
Is the information accurate, current and relevant?
Are uncertainty and limitations acknowledged?
What important information or context may be missing?
Which perspectives are represented?
Which perspectives should I investigate further?
Do credible sources disagree?
Can I explain why I trust this information?
Have I reached my own conclusion?
Check that:
I verified significant claims using authoritative sources.
I confirmed that every quotation, statistic and reference is genuine.
I consulted and cited original sources rather than relying on AI summaries.
I investigated errors, omissions, bias and competing perspectives.
I used evidence appropriate to my DP subject.
I distinguished facts from interpretations and opinions.
I developed my own analysis and conclusions.
I identified and referenced any AI-generated material included.
I can explain how the evidence supports my argument.
I take responsibility for the accuracy of the final work.
Remember: AI can help you begin an investigation, but it should never be the end of your research. Question the response, verify the evidence and use your own disciplinary judgement to reach a defensible conclusion.