By Alexa Amundson, Founder of BlackRoad OS
March 2026
Every school district in America is asking the wrong question about AI.
They're asking: "Should we ban AI in schools?"
The right question: "How do we build AI that makes students smarter instead of lazier?"
Because the ban isn't working. 69% of high school students use ChatGPT for homework. 92% of university students use AI tools. Banning AI in schools is like banning calculators in the 1970s — a temporary resistance to an irreversible change.
The change already happened. The question is whether we adapt intelligently or pretend we can turn back the clock.
Right now, AI in education looks like this:
1. Student gets homework assignment
2. Student opens ChatGPT
3. Student types the assignment into ChatGPT
4. ChatGPT produces a complete answer
5. Student copies the answer
6. Student submits the answer
7. Teacher grades it
8. Student learned nothing
This is the worst possible outcome. Not because AI is bad — because the AI being used wasn't designed for education. ChatGPT is designed to give the best answer as fast as possible. That's the opposite of what education needs.
Education needs productive struggle. The reaching for understanding. The moment where the brain works hard, fails, adjusts, and tries again. That's where learning happens. And ChatGPT short-circuits it completely.
Imagine the same scenario with Roadie:
1. Student gets homework assignment
2. Student opens Roadie
3. Student says "I don't understand this problem"
4. Roadie asks: "What part of the problem makes sense to you?"
5. Student explains what they know
6. Roadie asks: "Good. Now what if we applied that to the next step?"
7. Student thinks. Tries something. Gets it wrong.
8. Roadie asks: "Interesting approach. What happens when you check it with a simpler number?"
9. Student tries a simpler number. Sees the mistake. Fixes it.
10. Student solves the problem themselves.
11. Roadie says: "You just used algebraic substitution to verify your answer. Most students skip that step. That's real mathematical thinking."
Same starting point. Completely different outcome. The student did the thinking. Roadie asked the questions. And Roadie's specific, targeted praise reinforced the exact cognitive skill the student just practiced.
This isn't hippie pedagogy. It's backed by decades of cognitive science:
Bloom's two-sigma problem (1984): Students who receive one-on-one tutoring perform two standard deviations above students in traditional classrooms. That's the difference between the 50th percentile and the 98th percentile. Socratic tutoring is the gold standard.
Desirable difficulty (Bjork, 1994): Learning that feels harder in the moment produces better long-term retention. The struggle IS the learning. Making things easy (giving answers) feels good but doesn't stick.
Testing effect (Roediger & Karpicke, 2006): Retrieving information from memory strengthens the memory more than re-studying the information. When Roadie asks questions instead of giving answers, the student practices retrieval.
Self-explanation effect (Chi, 2000): Students who explain their reasoning learn more than students who receive explanations. Roadie's "walk me through your thinking" prompts activate this mechanism.
Zone of proximal development (Vygotsky, 1978): Learning happens best when the challenge is just above the student's current ability — hard enough to stretch, not so hard they give up. Roadie's scaffolded questions dynamically adjust to stay in this zone.
AI tutoring that gives answers violates all five of these principles. AI tutoring that asks questions activates all five.
Schools don't need to ban AI. They need AI that's designed for learning:
Teacher dashboard. Real-time visibility into every student's learning journey. Not test scores — process data. How many attempts? Where did they get stuck? What misconception is holding them back? When did the breakthrough happen?
Roadie provides this. Every interaction is logged (with student privacy controls). The teacher sees mastery heatmaps, intervention alerts, and concept progression for every student. One glance replaces an hour of assessment.
Parent portal. Secure view showing what was practiced, time spent, and growth streaks. No access to conversation content (students need private space to be wrong). Just enough visibility for parents to understand and support.
"Show your work" verification. Every Roadie session produces a timestamped, RoadChain-verified learning artifact. The student didn't just submit a correct answer — they submitted proof that they worked through the problem. Teachers can see the thinking, not just the result.
Anti-cheating by design. Roadie doesn't give answers in Socratic mode. There's nothing to copy. The student can't cheat because the AI refuses to short-circuit the learning process. The design prevents cheating rather than detecting it.
Subject adaptation. Math tutoring is different from history tutoring is different from creative writing tutoring. Roadie adapts its Socratic approach per subject. In math, it asks for numerical verification. In history, it asks for evidence. In creative writing, it asks for alternatives.
Roadie's badge system isn't gamification. It's a verified learning transcript.
Traditional transcripts show grades — outcomes. Roadie's badges show process:
Each badge is RoadChain-verified. Not a sticker from a teacher — a cryptographic proof of learning that can be shared with colleges, employers, or anyone who wants to see evidence of capability.
Imagine a college application that includes: "This student completed 200 Socratic tutoring sessions in calculus over 8 months, achieving Concept Conqueror status with verified mastery across all units." That's more meaningful than a letter grade.
Private tutoring costs $40-80 per hour. Three sessions per week costs $480-960 per month. $5,760-11,520 per year.
The kids who get this tutoring have an enormous advantage. Not because they're smarter — because they have a patient human who knows their learning style, remembers their struggles, and scaffolds their understanding.
Roadie provides the same experience for every student on BlackRoad OS.
The wealthy family's kid has a private tutor who remembers them? So does every other kid. The difference is Roadie costs $10-20/month instead of $500/month. And Roadie never gets tired, never has a bad day, never judges, and never forgets.
Is it as good as the best human tutor? Probably not. The best human tutors bring something AI can't replicate — yet.
Is it better than no tutor? Infinitely. And for the 80% of families who can't afford private tutoring, "better than nothing" is "everything."
Roadie doesn't replace teachers. It makes them superhuman.
A teacher with 30 students and no AI support:
A teacher with 30 students, each using Roadie:
That's not replacement. That's amplification. The teacher becomes the coach, the mentor, the facilitator of higher-order thinking. Roadie handles the drills.
If you're a teacher: try Roadie with one student for one week. Watch what happens when AI asks questions instead of giving answers.
If you're a parent: give your kid Roadie instead of ChatGPT for homework. Watch the difference between a child who copies an answer and a child who discovers one.
If you're a school administrator: stop debating whether to ban AI. Start asking which AI actually teaches. Roadie is the answer.
If you're a student: this is the friend who sits next to you and asks "but why does that work?" — the question that annoys you in the moment and teaches you for life.
The patient friend in the passenger seat.
Roadie — AI tutoring that makes students smarter, not lazier.
tutor.blackroad.io
Remember the Road. Pave Tomorrow.