TL;DR: Singapore universities are replacing traditional essay grading with AI-era thinking assessments that evaluate reasoning, adaptability, and prompt-engineering skills rather than prose style. This guide shows you how to prepare for, approach, and excel in these new assessments.
Step 1: Understand the New Assessment Format
AI-era thinking assessments typically present a complex, ambiguous problem (e.g., “Design a policy for autonomous vehicle liability”) and require you to produce a structured response in a live digital environment. You are often allowed—sometimes required—to use generative AI tools, but your grade depends on your prompt chain, your critical edits, and your final synthesis. The rubric scores “evidence of iterative reasoning” and “ability to detect AI hallucination,” not grammar or word count.
Step 2: Practice Prompt Engineering as Thinking
Before the test, write 10 multi-step prompts that force the AI to show its workings. For example, instead of “Write an essay on climate policy,” use “List three conflicting economic models for carbon pricing, then evaluate each against Singapore’s land constraint, then propose a hybrid model with trade-offs.” In the assessment, your first prompt should be broad, but your second and third must narrow based on the AI’s output. Treat every AI response as a draft to challenge, not an answer to accept.
Step 3: Build a “Verification Habit”
The core skill is spotting false logic or invented data. After each AI output, ask yourself: “Does this claim have a source I can check?” “Is this number plausible?” Type a follow-up prompt: “Show the calculation for this statistic, and flag any assumption you made.” Your on-screen log of these verification prompts is part of the grade. Never submit a final answer without at least one explicit “identify your own errors” prompt.
Step 4: Structure Your Final Response as a Decision Memo
Ignore essay paragraphs. Instead, use three sections: (1) Problem Restatement in one sentence, (2) Three Options with Pros/Cons in bullet form, (3) Recommended Choice with a “what would change my mind” caveat. This mirrors how executives think with AI assistants. Keep total length under 400 words—brevity signals clarity, not lack of depth.
Step 5: Manage Time in Three Phases
Allocate 20% of time to initial prompting, 50% to iterative refinement and verification, and 30% to final synthesis. If you get stuck, prompt the AI for “the weakest point in my current reasoning” rather than asking for more content. End with a one-sentence “limitation statement” about your own answer—this earns top marks for metacognition.
FAQ
Q: Can I use any AI tool during the assessment?
A: Yes, but the proctoring software logs every prompt and edit. Using AI to generate the final answer verbatim without critical intervention earns a zero—the goal is to show your thinking, not the AI’s.
Q: What if I’m not good at writing prompts?
A: Practice with free tools like ChatGPT or Claude for 15 minutes daily. Focus on asking for “trade-offs” and “counterarguments” rather than “facts.” Prompt skill is learnable and is now a graded competency.
Q: Will this replace all essays in university?
A: Not yet—it’s used in capstone projects, ethics modules, and policy courses. Traditional essays still exist for humanities, but you’ll see a 30-50% shift toward AI-assisted thinking tasks by 2026. Adapt early.
Leave a Reply