mirror of
https://github.com/rohitg00/ai-engineering-from-scratch.git
synced 2026-10-02 01:54:39 +08:00
Rohit Ghumare
dda194f840
Every "Test Your Understanding" quiz placed the correct answer in option B.
Across the 2026 questions in 338 quiz files the correct answer sat at index 1
in 61.5% of cases (uniform would be ~25%), and 107 files had every answer at B,
making the quizzes guessable without reading them.
scripts/debias_quizzes.py rewrites each question's option order with a
deterministic, content-seeded permutation and updates the correct index to
follow the moved answer. It is idempotent: options are canonicalised to a sorted
base before permuting, so re-running produces byte-identical output. Questions
whose options reference each other by position ("all of the above", "both A and
B") are left untouched. The correct-answer value, the option set, and every
explanation are preserved exactly; only order and the index change.
Result: A 23.8% / B 26.3% / C 23.5% / D 26.4%.
The script doubles as a CI guard: `--check` exits non-zero if any quiz is not
de-biased, wired into the curriculum workflow so new lessons cannot regress.
Fixes #368
79 lines
2.8 KiB
JSON
79 lines
2.8 KiB
JSON
{
|
|
"lesson": "24-regulatory-frameworks-eu-us-uk-korea",
|
|
"title": "Regulatory Frameworks - EU, US, UK, Korea",
|
|
"questions": [
|
|
{
|
|
"stage": "pre",
|
|
"question": "When did the EU AI Act enter into force, and what category applied first?",
|
|
"options": [
|
|
"August 2025; all categories applied at once",
|
|
"January 2024; high-risk systems applied that day",
|
|
"August 2024; prohibited practices and AI literacy applied 2 February 2025",
|
|
"August 2026; only GPAI applied at force-date"
|
|
],
|
|
"correct": 2,
|
|
"explanation": ""
|
|
},
|
|
{
|
|
"stage": "check",
|
|
"question": "What is the maximum monetary penalty under the EU AI Act applicable from 2 August 2026?",
|
|
"options": [
|
|
"Up to 5M EUR or 1% of global turnover",
|
|
"Up to 1M EUR or 0.1% of global turnover",
|
|
"Up to 15M EUR or 3% of global turnover",
|
|
"Unlimited"
|
|
],
|
|
"correct": 2,
|
|
"explanation": ""
|
|
},
|
|
{
|
|
"stage": "check",
|
|
"question": "Which three chapters make up the GPAI Code of Practice (10 July 2025)?",
|
|
"options": [
|
|
"Capability, Control, Compliance",
|
|
"Transparency, Copyright, Safety and Security",
|
|
"Privacy, Bias, Watermarking",
|
|
"Training, Inference, Deployment"
|
|
],
|
|
"correct": 1,
|
|
"explanation": ""
|
|
},
|
|
{
|
|
"stage": "check",
|
|
"question": "What does the February 2025 UK AISI rebrand to 'AI Security Institute' signal?",
|
|
"options": [
|
|
"A narrower scope focused on frontier-capability security, dropping algorithmic bias and free-speech framings",
|
|
"Closure of UK frontier evaluation",
|
|
"Merger with the EU AI Office",
|
|
"Expansion into algorithmic bias"
|
|
],
|
|
"correct": 0,
|
|
"explanation": ""
|
|
},
|
|
{
|
|
"stage": "post",
|
|
"question": "What does the US CAISI rebrand (June 2025) signal in policy direction?",
|
|
"options": [
|
|
"Mandatory open-source release of frontier models",
|
|
"Merger with FDA",
|
|
"A shift toward 'pro-growth AI policies,' with emphasis on standards and innovation support and reduced emphasis on pre-deployment evaluation",
|
|
"Stricter pre-deployment safety evaluation"
|
|
],
|
|
"correct": 2,
|
|
"explanation": ""
|
|
},
|
|
{
|
|
"stage": "post",
|
|
"question": "What is the headline obligation in Korea's AI Framework Act (effective January 2026)?",
|
|
"options": [
|
|
"Identical adoption of the EU AI Act",
|
|
"Total prohibition of foreign AI providers",
|
|
"Mandatory open-weights for all generative models",
|
|
"Article 12 establishes an AISI under MSIT and mandates local representatives, risk assessment, and safety measures for high-impact and generative AI"
|
|
],
|
|
"correct": 3,
|
|
"explanation": ""
|
|
}
|
|
]
|
|
}
|