You're pledging to donate if the project hits its minimum goal and gets approved. If not, your funds will be returned.
$15,000 left to reach the funding goal.
Socra is a Socratic AI tutor that asks questions instead of handing over answers. It tracks which concepts a learner has mastered, schedules review, and runs weekly evaluations on whether its own hints made the learner more or less dependent on AI. I built the first version for my own path through Harvard's CS50 and it runs today inside my company's app. The question I care about: as AI gets better, do people get sharper or more passive? Most tutoring products optimize for answers delivered. Socra optimizes for what the learner can do without it the next day.
Goal: measure whether a tutor designed to withhold answers reduces AI dependence in adult learners returning to school. Approach: run Socra with a small study of 20 adult learners over 6 months, using the weekly dependence evaluations built into the product, and publish the results and method openly.
Learners show higher unaided performance on previously covered concepts at the next session, and the weekly evaluations show hint dependence falling over time. I will publish a public write-up of the results, including negative findings.
$5,000 to $15,000 for 6 months of inference and evaluation costs, running the 20-learner study, and the public write-up.
I am a solo founder. I am a first-generation college student from a household earning under $30,000. In 2017 my dad's trucking company failed and I left college to support my younger sister and brother; I have not been enrolled since. Without a degree I sold enterprise deals at BMW ($9.8M closed against a $6M quota, 2021 to 2024), then worked in identity verification and fraud prevention at Prove and as Founding Account Executive at Heka Global (a past role). I now build Tounkara Group full time, write the automations myself, and am taking Harvard's CS50 on my way to a Harvard Extension degree.
Likely causes: too few learners complete the study, or the effect is too small to detect. Outcome: a smaller public write-up of what I learned, with the costs tracked against the grant.
$0. Tounkara Group is bootstrapped with no outside funding.