You used SPSS for your statistics. Nobody questioned it. You used Excel. Fine. Now you use an AI-powered tool, and suddenly there is a debate. The question keeps coming up in student forums, in emails to supervisors, and in the DMs we receive every week: am I allowed to use AI for my thesis statistics? The answer has less to do with technology than it does with policy. The computational methods are the same either way.
The Short Answer
For statistical computation, yes. It is almost universally allowed. Universities have always expected students to use statistical software. SPSS, R, Stata, Python, SAS, jamovi, JASP. None of these are controversial. An AI-powered statistical tool that runs the same tests (t-tests, ANOVA, chi-square, regression) using the same algorithms (scipy, statsmodels, or R's base libraries) falls in the same category. The math does not change because the interface is different.
The real distinction is between AI as a computation tool and AI as a writing tool. Computation is allowed. Writing, in most academic contexts, is restricted or outright prohibited unless you disclose it fully. If a tool runs a Shapiro-Wilk test on your data and returns W = 0.94, p = .18, it has done the same thing SPSS does. The fact that AI helped select the test or format the output does not change the mathematics.
Think of it this way. Your calculator does not get accused of academic dishonesty when you use it for long division. A statistical tool that automates a Mann-Whitney U test is the same kind of thing, just several steps further along the automation spectrum.
Where Universities Draw the Line
Most universities published AI policies between 2023 and 2025. If yours has not, ask your faculty office directly. But across the policies we have read (and we read a lot of them, because our students send them to us), three categories appear consistently.
First, using AI to process, calculate, or analyze data is generally permitted. This includes running statistical tests, cleaning datasets, checking assumptions, and producing tables or charts. Second, using AI to generate written text and submitting it as your own is generally prohibited, or requires full and explicit disclosure. Third, using AI to check grammar, spelling, or to rephrase sentences you already wrote sits in a middle ground. Most institutions allow it with a brief mention in your methods or declaration.
UNESCO's 2023 guidance on AI in higher education recommended that institutions allow AI tools while requiring transparency about their use. A large number of European and US universities followed this model through 2024 and 2025. The Russell Group in the UK published principles in mid-2023 that explicitly stated students should learn to use AI tools as part of their education. The European University Association echoed similar positions. By 2026, the consensus is fairly stable: tools are fine, opacity is not.
The word that appears in almost every policy document is "transparency." Use tools, but say which ones and how you used them.
What Kind of AI Tool Makes a Difference
There is a meaningful gap between a general-purpose chatbot and a domain-specific computation tool, and universities are starting to recognize it.
A chatbot that "generates" statistical results from a text prompt is dangerous. It can fabricate p-values. It can invent test statistics that look plausible but have no computational basis. Ask a general-purpose AI to "run an ANOVA on my data" without actually giving it the data in a computable format, and it may produce a table of numbers that never existed. This is the scenario universities worry about, and rightly so. A student who copies fabricated statistics into a thesis has committed something worse than plagiarism. They have introduced false data into an academic work.
A statistical computation tool that takes actual data, runs actual algorithms, and returns real computed results is a different thing entirely. The output is deterministic and reproducible. Feed the same dataset twice, get the same answer. There is no generation involved. There is only calculation.
Academic Stats Agent was built with this distinction at its core. It runs scipy and statsmodels computations on your uploaded dataset. The AI component writes the prose around verified numbers, but a fail-closed guard prevents any fabricated statistic from reaching the report. Every number in the output is computed from your data, not generated from a language model's probability distribution. When the tool reports F(2, 47) = 4.32, p = .019, those values came from your actual dataset passing through an actual ANOVA computation.
How to Disclose AI Tool Use in Your Thesis
APA 7th edition includes updated guidance (from 2023 onward) on citing AI-assisted tools. The recommended approach: describe the tool, what you used it for, and when, in your Methods section. Keep it factual. One or two sentences is usually enough. (For guidance on formatting your statistical results in APA-7, see our guide to reporting p-values.)
Here is an example you can adapt for your own thesis Methods section:
"Statistical analyses were performed using Academic Stats Agent (StudentStats.net, 2026), an AI-powered statistical analysis platform that executes scipy and statsmodels computations on uploaded datasets and produces APA-formatted reports."
If your university has its own AI disclosure template, use that instead. Some institutions ask students to fill in a standardized form listing every AI tool they used, what they used it for, and on which dates. Others want a separate appendix with screenshots or logs. The format varies. The expectation of honesty does not.
Some universities also ask for a signed AI declaration as a separate appendix, alongside the standard plagiarism declaration. Check with your supervisor before submission. Better yet, check before you start your analysis. A two-sentence email explaining which tool you plan to use and how you plan to use it can save real problems at the defence stage. Supervisors appreciate being told in advance. They do not appreciate surprises during the viva.
What We Tell Students Who Ask Us
Over more than 300 projects, we have seen the anxiety around this question firsthand. Students email us worried that using any software with "AI" in the name will get them flagged. The concern is understandable but usually misplaced. The students who get in trouble are not the ones who use statistical software. They are the ones who submit chatbot-generated paragraphs as their own analysis, without disclosure, and sometimes without even checking whether the numbers in those paragraphs are real.
Our recommendation is straightforward. Use AI tools for what they are good at: computation, formatting, catching errors in your test selection, checking assumptions you might have missed. Then write the interpretation yourself. Your supervisor wants to see that you understand your results, not just that you can paste them.
A sentence from you that reads "the intervention group showed significantly lower anxiety scores (p = .012), which aligns with previous findings by Kowalski et al. (2021) and may reflect the six-week duration of the programme" is worth more than a perfectly formatted AI-generated paragraph that says the same thing with more syllables. Your voice matters. Your ability to connect the statistics to your research question matters. The computation itself is the easy part.
We have never had a student penalized for using a statistical computation tool and disclosing it properly. We have seen students penalized for pretending they did not use one. The difference comes down to honesty, and that has always been the rule in academic work, long before AI entered the picture.
Key takeaway: AI-powered statistical tools are generally permitted in academic work, just as SPSS and R are. The line is at AI-generated text submitted as your own writing. Disclose your tools in the Methods section and confirm the policy with your supervisor.
Frequently Asked Questions
Is using AI tools for thesis statistics academic misconduct?
Generally no. Most universities treat statistical software, including AI-assisted analysis tools, the same as SPSS or R: computational aids, not ghostwriting. Misconduct typically arises when AI writes text you submit as your own or when your institution requires disclosure and you fail to provide it.
Do universities detect AI-generated statistics?
AI detectors target generated prose, not calculations, so correctly computed statistics are not flagged. The real risk is the thesis defense: supervisors quickly notice when a student cannot explain the analysis in their own work, so you must understand and be able to justify every test you report.
Do I need to disclose AI tool use in my thesis?
Check your university's AI policy. Many institutions now ask students to name the software used for analysis in the methodology chapter, exactly as they would cite SPSS or R, and some require a short statement describing what the AI tool was used for.