Time vs Money: What "Free" ChatGPT Really Costs a student
- Samisa Abeysinghe
- 5 days ago
- 10 min read
Generic AI chatbots cost nothing to use. Idasara Academy charges a subscription. So why would anyone pay? Because for a student in an exam year, the scarcest resource is not money — it is time. Here is the honest accounting.
At a glance
A bare AI chatbot is free in rupees but charges you in hours: setting up chats, feeding it the syllabus, crafting prompts, and double-checking every answer is study time spent building a tool instead of learning.
Even used perfectly, a chat window cannot show you your improvement path. It answers and forgets. A study engine remembers every mistake and shows you where you are versus where you could be.
Prompt packs teach the art of prompting — a skill you keep for life — without the nightly chore of prompt crafting.
Hallucination is a built-in property of how language models work, not a rare glitch. Guardrails and a verified curriculum knowledge base exist precisely for this.
Plain ChatGPT is genuinely fine for some things — we say so below, honestly.
Every week, a student or a parent asks us some version of the same question:
"Why should I pay for Idasara when ChatGPT is free?"
It is a fair question. It deserves a fair answer — not a defensive one, and not a scary one. So let me try to answer it the way I would answer a friend.
The answer begins with changing the currency.
When we say ChatGPT is "free," we mean free in money. But money is not the only thing a student spends. A student in an exam year has, at most, a few good study hours in a day — after school, after travel, after meals, before sleep takes over. Those hours are the whole game. Every plan, every tool, every choice either protects those hours or quietly eats them.
So the honest comparison is not free versus paid. It is what does each path cost in hours, and what do those hours buy you?

What does "free" actually cost?
Here is what nobody tells you about studying with a bare, general-purpose chatbot. Before it teaches you anything, you must teach it.
Think about what it actually takes to turn a generic AI into a study tool for a Sri Lankan national exam:
You set up the workspace. New chats, projects, custom instructions — organized by subject, if you are disciplined about it.
You feed it the knowledge. The chatbot was trained on the whole internet, which means it defaults to foreign syllabi — US and UK curricula, different notations, different marking conventions. You must keep telling it, in every chat, what syllabus you are on.
You craft the prompts. Long setups to get the level right, the language right, the format right. Get the prompt slightly wrong and the answer is confidently off-target.
You rebuild the context, again and again. A new chat starts from zero. The careful setup you built last Tuesday is gone.
You verify everything. Every formula, every definition, every "fact" needs checking against your textbook — because you cannot hand in an answer the AI invented.
None of this is learning. It is tool-building. In simple terms: with a bare chatbot, you become the toolsmith — and the workshop hours come out of your study hours.
There is a name for part of this cost: the prompting tax. Every setup you type, every answer you double-check, every context you rebuild — that is time the tool charged you before it did any work.
Now the same task, side by side. Say you are stuck on a past-paper question tonight:
The task tonight | With a bare chatbot | With Idasara |
Get help at your syllabus level | Write a long setup prompt; hope it holds | Already grounded in the national curriculum — ask directly |
Trust the explanation | Cross-check against the textbook yourself | Answers come from a verified, curriculum-guarded knowledge base |
Practise properly | Ask it to invent questions; quality varies | Work revision questions and past-paper practice built for your syllabus |
Get your answer marked | Paste it in; feedback is generic and unverified | The exam-marking tool scores it the way an examiner would |
Learn from the mistake | The chat forgets it by tomorrow | The mistake goes into your Mistake Bank automatically |
Know where you stand | No way to know | Your Readiness Score moves — you can see it |
Both columns end with an answer. Only one of them ends with you knowing more about yourself than when you started.
What can a chat window never tell you?
This is the part that matters most, and it has nothing to do with prompt quality.
Suppose you did everything above perfectly. Perfect prompts, perfect setup, every answer verified. You would still be missing the one thing an exam-year student needs most: a mirror.
A general chatbot answers your question and forgets you. Tonight's mistake in algebra, last week's confusion about chemical bonding, the essay structure you keep getting wrong — each one vanishes into a separate chat log. The tool has no memory of your journey, so it can never tell you the two things that actually decide your result:
Where am I now? And where could I be?
Idasara is built around exactly that memory. Every error you make feeds your Mistake Bank — a living record of your personal weak spots. Every practice session moves your Readiness Score. Over months, the platform learns the shape of your preparation — not a generic student's — and shows you the gap between your current self and your possible self, and the shortest path across it.
Does that make sense? A tutor who forgot you after every single lesson would also be cheap. That has never been what a good teacher is for.
What do you get back? Time — and a plan for it
Here is the part of the accounting most comparisons skip: the hours you recover.
When the setup tax disappears and the checking tax disappears, your study hour becomes a full study hour. And because the platform knows your weak spots, it does not send you back through whole chapters you have already mastered — it points you at the gaps. Your study plan bends around your school days and your energy, not the other way around.
What are those recovered hours for? This is where I will tell you what we actually believe, because it is the reason Idasara exists at all.
The goal was never for you to study more. It is for you to study better, in less time — and to take the hours you win back and spend them on being a whole person. Sport. Family. Rest. Reading something that has nothing to do with the exam. Staring at the sky, if that is what restores you. We call this Me Time, and we consider it a feature — perhaps the feature.
A free tool that quietly consumes your evenings is not free. A paid tool that returns your evenings is not expensive.
Learn the art of prompting — skip the chore of prompt crafting
At this point a thoughtful student will push back: "But if I wrestle with ChatGPT myself, at least I learn prompting. Isn't that a life skill?"
Yes — and this objection deserves a careful answer, because there are two different things hiding inside the word "prompting."
Prompt crafting is the plumbing: the syllabus setups, the format instructions, the workarounds. It is specific to one tool, it dies with the chat window, and repeating it every night teaches you very little after the first week.
The art of prompting is something else entirely: asking a sharp question, giving the right context, judging whether the answer is sound, and pushing back when it is not. That skill transfers to every AI tool you will ever use, in university and in work. It is a genuine 21st-century literacy.
Idasara's Prompt Packs are designed to teach you the second thing without wasting your year on the first. Each pack is a structured set of prompts for the topic you are studying: you answer on paper first, then take what is left to Ask AI — and in doing so you learn, by guided practice, how to talk to an AI properly and how to evaluate what comes back. The art, not the chore.
What about the answers themselves?
Now the uncomfortable subject — the one every parent should ask about.
Language models sometimes state false things with complete confidence. This is not a rare bug that will be patched next month; researchers describe hallucination as a consequence of how these models are built and trained — a feature of the design, not a defect in it. Models also carry the biases of the data they were trained on, most of which was not written for a Sri Lankan classroom. For a professional double-checking a draft, that is manageable. For a sixteen-year-old memorizing a confidently wrong formula three weeks before a national exam, it is a real cost — paid later, at the worst possible time.
Let me be clear about something, because it matters for honesty: Idasara is not anti-AI. Idasara runs on AI. The same underlying technology powers our Ask AI. The difference is what stands between the model and your child:
Grounding. Ask AI answers from a verified knowledge base built on the national curriculum — not from the open internet's average opinion.
Guardrails. The system is constrained to your syllabus, your level, your language — by design, not by a prompt you had to remember to type.
Taught skepticism. We teach students how to spot when an AI is wrong — because the judgment to check the machine is a skill you will need long after the exam. This is the same guided-use principle that UNESCO's guidance on generative AI in education urges schools worldwide to adopt: not banning the technology, and not handing it over unguarded either.
Guardrails do not make any AI perfect. They make it accountable — and they mean the checking is done by the system and the teacher-designed knowledge base, not squeezed out of your child's study hours.
When is plain ChatGPT enough?
An honest comparison names the cases where the free tool wins. Here they are.
If you are exploring out of curiosity — a hobby, general knowledge, a topic with no fixed syllabus and no exam at the end — a general chatbot is a marvellous companion. If you are an adult learner with time to spare, comfortable double-checking claims, you may genuinely enjoy building your own study system; you will learn things from the building itself. And for quick, low-stakes questions where a wrong answer costs you nothing, free is exactly the right price.
We did not build Idasara for those moments. We built it for the student who has one exam year, a few good hours a day, and no spare evenings to be a toolsmith.
Spend money, or spend time — but spend on purpose
So, the honest summary of the whole comparison:
With a bare chatbot, you pay in time — setup, prompting, checking — and at the end of the year, you hold a stack of chat logs and no map of how far you came. With Idasara, you pay in money, and the system pays you back in hours, in a visible improvement path, and in an AI literacy you will carry into whatever comes after school.
If you are a parent reading this, the question underneath the question is probably: what am I actually buying? You are not buying answers — the internet is full of free answers. You are buying back your child's evenings, a guarded space to use powerful technology safely, and a window into their real progress instead of a hopeful guess.
Neither path is wrong. But choose deliberately — because you will spend one currency or the other, and only one of them comes back.
Idasara starts free, so you can test everything this article claims before a rupee changes hands: academy.idasara.org. Try it properly for two weeks — the loop, the marking, a prompt pack. Then look at what your Readiness Score already knows about you, and ask whether a chat log ever could.
Your first saved hour is yours. Spend it on something that makes your eyes glitter.
Frequently asked questions
Is ChatGPT good enough to study for O/L or A/L?It can explain concepts well, but it defaults to foreign syllabi, sometimes invents formulas, and keeps no record of your mistakes or progress. To use it safely for a national exam you must supply the syllabus context, craft careful prompts, and verify every answer — a real, recurring time cost that comes out of your study hours.
Is Idasara against students using AI?No — Idasara is built on AI. The difference is that its Ask AI is grounded in a verified national-curriculum knowledge base with guardrails, and the platform explicitly teaches AI literacy: how to prompt well and how to judge what the AI returns.
Does Idasara completely eliminate AI hallucinations?No AI system can honestly promise that. Research indicates hallucination is inherent to how language models work. Idasara reduces the risk with curriculum grounding and guardrails — and teaches students to verify answers, because that judgment is itself an exam-relevant and career-relevant skill.
What does "the prompting tax" mean?The recurring time a bare chatbot charges before it helps you: writing setup prompts, restating your syllabus and level, rebuilding context in every new chat, and double-checking outputs. Idasara removes it by being pre-grounded on the national curriculum.
What does Idasara cost?It starts free, so you can try the loop, the marking tool, and a prompt pack before paying anything. Current plans are on the pricing page.
Will I still learn prompting if Idasara does the setup for me?Yes — more deliberately, in fact. Prompt Packs teach the transferable art (sharp questions, right context, critical evaluation) through guided practice, instead of the tool-specific chore of nightly prompt crafting.
Sources & further reading
All links verified live on 2026-08-19.
Why Language Models Hallucinate — Kalai, Nachum, Vempala & Zhang (arXiv, September 2025) — the research argument that hallucination follows from how models are trained and evaluated.
AI hallucinations are a feature of LLM design, not a bug (Nature, 2025) — why confident wrong answers are inherent, not incidental.
UNESCO — Guidance for generative AI in education and research — the global case for guided, rather than banned or unguarded, use of generative AI in schools.
Idasara Academy — the platform discussed in this article; pricing and why Idasara.
DigiReady: Talk to AI properly and DigiReady: Spot when AI lies — the free lessons behind the AI-literacy claims above.
The Gen AI Evolutionary Journey: 2017 to 2035 — our companion piece on why judgment, not tool access, is the scarce skill.




Comments