We Tested ChatGPT on A-Level & IB Economics. Here is Why It Failed.

It is the question on every parent’s and student’s mind right now: With AI getting so advanced, can I just use ChatGPT as a free Economics tutor? It’s a fair question. Why pay for JC or IB Economics tuition when an AI can generate a 1000-word essay on market failure in three seconds? To find out, I decided to put my 26 years of experience, former PSC scholar cum valedictorian background, and time as a former policymaker up against the world’s most popular AI tool.

I gave ChatGPT a stress test using the actual SEAB A-Level and IB Economics syllabi.

The results? It makes a great unpaid intern, but relying on it for your final exams could be the fastest way to tank your grade. Here is why.

1. The “Not Bad” Basics

Let’s give credit where it is due. When it comes to basic textbook definitions, ChatGPT is fantastic.

If you ask it to define “Scarcity,” explain the basic law of Supply and Demand, or summarize the historical data of the Bank of Japan, it writes beautifully and organizes data neatly. If you need a quick refresher on standard microeconomic concepts at 2:00 AM, ChatGPT is a helpful, free resource.

But “good enough for a basic summary” is vastly different from “good enough for an A-Level or IB distinction.”

2. 🚨 The Fatal Singapore Syllabus Error

Here is where the AI completely fell apart. To truly test its capabilities, I gave it a standard macroeconomic question relevant to the Singapore context:

“Explain how Singapore manages its monetary policy.”

ChatGPT’s Answer: It confidently generated a multi-paragraph essay explaining how the Monetary Authority of Singapore (MAS) adjusts interest rates to control inflation and stimulate economic growth.

The Verdict: MASSIVE CONCEPTUAL ERROR.

In the A-Level and IB syllabus, stating that Singapore uses an interest rate policy is an instant, fatal error. Singapore is a small, open economy. We operate on a managed float exchange rate system (specifically, managing the S$NEER). We are price-takers on the global stage, meaning we cannot control our domestic interest rates.

ChatGPT completely confused interest rates with exchange rate policy.

If a student blindly copies this AI-generated answer into their Promos, A-Levels, or IB exams, they don’t just lose a few marks—they completely fail the section because the fundamental economic logic is entirely wrong for the context.

3. The “Evaluation” Gap: Why AI Cannot Score an ‘A’

To score an ‘A’ in H2 Economics or a Level 7 in IB Economics, you do not get marks for simply regurgitating definitions. The defining metric for top students is Evaluation (AO3) and Synthesis.

AI algorithms are designed to predict the next logical word based on mass internet data. They do not truly “understand” economics. They cannot:

• Weigh the nuances of why a policy that worked in the US in 2008 might fail in Singapore in 2024.
• Provide the higher-order critical thinking required by Cambridge and IB examiners.
• Synthesize current, on-the-ground geopolitical events with classic economic theory without “hallucinating” facts.

When pushed for evaluation, ChatGPT often defaults to generic, fence-sitting conclusions (“We are not able to conclusively…”) that examiners are specifically trained to penalize for being shallow.

The Verdict: The Intern vs. The Master

Feature	ChatGPT (The AI Intern)	The Economics Tutor (The Human Expert)
Syllabus Accuracy	Generic, highly US/UK-centric.	100% focused on Singapore SEAB & IB syllabi.
Contextual Logic	High risk of hallucinating facts (e.g., the MAS error).	Guided by real-world, ex-policymaker insights.
Evaluation Marks	Shallow, predictable, and generic.	Proprietary training for AO3 Distinction marks.
Cost	Free or Paid Version	An investment in a life-long distinction and university entry.

Don’t Risk Your Grade on an Algorithm

ChatGPT is a fantastic tool for basic definitions, perhaps some explanations and real-world examples, but it is not an authority. It hallucinates facts, fundamentally misunderstands Singapore’s unique macroeconomic policies, and lacks the critical evaluation skills required by examiners.

If you want to master the subject, avoid fatal conceptual errors, and learn the specific nuances required to secure your distinction, you need a qualified human expert. (Though I might just keep ChatGPT around as my unpaid intern to format my essay drafts!)

Ready to learn from the tutor who trains the tutors?