Can AI validate scientific claims, without a statistician in the loop?
That’s not just a question. It’s the crisis of our time in the age of LLMs.
As AI floods research with machine-generated hypotheses, the bottleneck isn’t in creation it’s in validation. Manual statistical testing is too slow, too exclusive, and not scalable.
That’s exactly what I’ve set out to solve.
Introducing: A Framework for Automated Hypothesis Testing
DOI: 10.5281/zenodo.16054306
What it does:
Accepts hypotheses in natural language
Uses transformer-based NLP to parse statistical intent
Automatically selects & executes the right statistical tests
Summarizes results in plain English via LLMs
Think SPSS × ChatGPT but built for open science, reproducibility, and scale.
Why now?
Unlike agentic systems like POPPER (Stanford), which are powerful but complex, this framework is modular, lightweight, explainable, and made for non-experts too like educators, students, policymakers, and independent researchers.
I am looking to collaborate with:
Universities & research labs
AI/ML developers
Citizen science platforms
EdTech innovators
If you’re working on democratizing science, automating data workflows, or enabling reproducible research let’s talk.
Interested?
Fill this form to explore collaboration: https://app.nocodb.com/p/researchinterest
Or read the preprint here: A FRAMEWORK FOR AUTOMATED HYPOTHESIS TESTING
DM me or reach out at i@hardiktiwari.com or https://linkedin.com/in/itiwarihardik/
Would you use this in your research workflow? Comment below or tag someone who should see this.
Empowering Open Education Through Automated Scientific Reasoning and Hypothesis Testing
In the spirit of open education, This Framework for Automated Hypothesis Testing breaks down the technical barriers that often exclude learners from engaging with real-world research. By enabling students, educators, and independent scholars to formulate hypotheses in plain language and receive statistically valid results without requiring advanced training, the system transforms passive learning into active scientific exploration. It serves as an AI-powered research assistant bridging statistics, critical thinking, and computational literacy while promoting reproducible research practices. This aligns with the core mission of open education (atleast according to me): to make high-quality, interactive, and participatory learning accessible to all.