The machine may be right. But the signature is yours.
Mayéutica teaches students to reason through real situations and use artificial intelligence without believing it blindly. The machine may be right, wrong, challenge them or agree with them. The final decision belongs to the student, who must support it with a reason.
The machine may be right. But the signature is yours.
Reason before calculating
Mathematics is the terrain, not the objective. Each task starts from the exercise the teacher was already going to assign and turns it into a real situation. The student moves through three moments, always in the same order: logic, learning and mathematics.
- Logic. First decide what is happening, without numbers.
- Learning. Then learn a brief, verified fact about the world.
- Mathematics. Only at the end does calculation begin. The system asks backwards and uses rescue steps, but never gives the result.
The contradictor
During the logic step, a machine comments on the student's decision. Before the task begins, the student is told that this machine is sometimes right and sometimes wrong, sometimes agrees and sometimes disagrees. It never reveals which case is happening.
- The student is right and the machine agrees: continue.
- The student is right and the machine challenges them with a convincing but false argument: hold the position.
- The student is wrong and the machine challenges them correctly: change the position.
- The student is wrong and the machine agrees: discover that validation by a machine does not make an answer true.
There is always a reveal at the end. The purpose is not to confuse the student but to train a specific habit: listen to AI, evaluate what it says and make the final decision independently.
Use AI to understand, not to skip
Mayéutica does not try to prevent students from opening ChatGPT, Gemini or another tool. It may invite them to ask elsewhere, but the student must explain the problem in their own words. When returning, they type what the other system said and decide whether that answer holds up. Searching is part of learning; the query does not replace reasoning.
One real task, end to end
In the direct-proportion demonstration, the teacher starts with a tap that loses three quarters of a litre every two hours. The system proposes real situations and the teacher corrects them because the goal is for students to decide when a relationship is proportional and when it is not. A rain scenario on a dirt road is selected and frozen for the whole class.
The student first decides whether the observed rainfall can be projected. The contradictor argues back. The student then learns what one millimetre of rain means. Only after that do the original numbers return and the mathematics begins. If the student treats the amount collected in two hours as if it were the amount for one hour, the system recognises that difficulty and steps backwards with questions.
A path the teacher can review
Mayéutica does not return only right or wrong. The teacher can see what the student decided, whether a correct answer was maintained when challenged, whether an incorrect answer changed, whether false validation was believed, where the mathematics became difficult and how many rescue steps were needed.
Across the class, the system can surface something more useful than a grade: for example, that 17 of 28 students treated a two-hour quantity as a one-hour quantity, or that 9 of 28 believed the contradictor when it presented a false argument. That tells the teacher what may need to be taught again.
Why now: data with source and date
- 52% use AI chatbots at least once a week for learning. Only 10% report never using them for schoolwork. Source: PISA 2025, OECD. Published: September 2026.
- Among AI users, students who had been taught to evaluate the quality of AI output performed better than those who had not. The OECD describes an association, not a causal finding. Source: PISA 2025, OECD. Published: September 2026.
- 8 in 10 Argentine 15-year-olds do not reach the minimum mathematics level. Source: PISA 2025, OECD. Published: September 2026.
- In sixth grade, 5 in 10 reach the expected level in mathematics. Source: Aprender 2025. Results published: June 2026.
- In the final year of secondary school, 14.2% reach a satisfactory mathematics level and 54.6% remain below basic. Source: Aprender 2024. Results published: June 2025.
- 55% report classmates being distracted by devices in almost every class, compared with a 28% OECD average. Source: PISA 2025, OECD. Published: September 2026.
- Only 6 in 10 children can use a computer at home for homework; in lower socioeconomic households, 3 in 10. Source: survey cited by MDZ. Published: February 2026.
Mobile first, short sessions
The experience is designed for the first two years of secondary school and for the device students actually have at hand. A task lasts ten to fifteen minutes, works on small screens and can resume after a connection drop. It avoids notifications, sounds and reward animations that compete for attention.
Facts, not profiles
The system reports measured facts, never personal traits. It may say that a student struggled eight out of ten times when moving from fractions to decimals. It may not label that student slow, brilliant or incapable. There are no rankings and students never see another student's position.
The teacher keeps authority
The content begins with the teacher. The teacher uploads an exercise, corrects the proposals, chooses one situation and defines its pedagogical purpose. The selected task is frozen for the class. Mayéutica makes visible how each student reasoned and where the class struggled; grading and pedagogical judgement remain human.
What it is not
- It is not an AI detector.
- It is not a homework solver.
- It is not a virtual classroom.
- It does not diagnose or classify students.
- It does not prohibit digital tools.