Papers and technical notes
What we have written about how DuckyHelper teaches, and why.

It cuts both into twelfths on the board, so the answer is right there.These are DuckyHelper's own white papers and notes. They explain how DuckyHelper is designed and which published research each design choice comes from. They are white papers from MingLLM, Inc., the company that makes DuckyHelper: not peer reviewed, and they report no experiments or results of our own.
Built to Teach
How ten findings about how people learn, from Bloom's two sigma result to the risks of unguarded AI help, each became a teaching rule or a tool in DuckyHelper, and where the principles pull against each other.
Read the paperShowing, not telling
Why DuckyHelper's app has a pen and a pointer on the learner's real screen, how it decides where a mark goes, what it does instead in the app's own tutor tab, and the privacy limits on both.
Read the paperHints before answers
What the evidence says about AI help that hands over answers, and the guardrails in DuckyHelper: a hint ladder that leaves the answer for last, feedback that marks where without fixing, and what the guardrails cannot do.
Read the paperThe annotated bibliography
Every paper in the library, 100+ of them, with its full citation and our one-line note, in one PDF.
What these are White papers, not peer reviewed. They describe design and cite published research. They contain no study of DuckyHelper and no numbers of our own.
Questions
Are these papers peer reviewed?
No. They are white papers written by the team at MingLLM, Inc., which makes DuckyHelper. They describe the design and cite peer-reviewed research, but they have not been through peer review themselves.
Who wrote them?
The DuckyHelper team at MingLLM, Inc., the company that makes DuckyHelper. They are written in plain words for parents, teachers and researchers, and every finding links to its source.
Can I cite them?
Yes. Each paper page has a citation block. Please cite them as white papers from MingLLM, Inc., and cite the original research for any finding they mention.
Keep exploring
Library
100+ real learning-science papers behind DuckyHelper, sorted by principle, each with a citation, a link to the publisher and a one-line summary.
OpenBuilt to teach
DuckyHelper is trained on 1,000,000+ tutoring examples and inspired by 100+ learning-science papers: it asks you back instead of handing you the answer.
OpenPrinciples
Twelve learning-science principles behind DuckyHelper, from one-on-one tutoring to spacing, each with the research and what DuckyHelper does because of it.
Open
Do anything
Ask it a question or hand it a job, in any app.
Doing jobs for you is in the Mac app. Asking works on Mac, Windows and Linux.
