Four questions on AI history, terminology, and the spectrum of AI capability. No pressure — answers are explained either way.
Question 1 of 4
Question 1 of 4
Which analogy best describes the relationship between AI, Machine Learning, and Deep Learning?
AThree separate technologies that occasionally overlap
BRussian dolls — Deep Learning sits inside Machine Learning, which sits inside AI
CA timeline showing what came before what
DThree competing schools of thought
Explanation
Every Deep Learning system is a Machine Learning system, which is an AI system. But not every AI system uses ML, and not every ML system uses Deep Learning. Concentric circles — or Russian dolls — are the right mental model. The relationship is one of containment, not competition or chronology.
Question 2 of 4
What converged after 2012 to trigger the modern AI explosion?
AGovernments invested heavily in national AI programmes
BScientists discovered fundamentally new mathematics
CLarge labelled datasets, affordable GPUs, and deep learning architectures all matured together
DThe internet was invented, creating new data sources
Explanation
The ideas behind deep learning existed for decades. What changed was the convergence of three things: ImageNet and similar large labelled datasets, GPUs that could parallelise matrix operations cheaply, and architectural breakthroughs like AlexNet in 2012. No single ingredient was sufficient — all three were required.
Question 3 of 4
Alan Turing's "imitation game" proposed that a machine could be called intelligent if it could:
ASolve any mathematical problem faster than a human
BPass as human in a blind text conversation
CBeat a grandmaster at chess
DLearn from data without being explicitly programmed
Explanation
In Turing's test, a human judge conducts text conversations with a machine and a human simultaneously without knowing which is which. If the judge cannot reliably tell them apart, the machine has passed. Turing deliberately chose behaviour — not internal experience — as the criterion, sidestepping the question of consciousness entirely.
Question 4 of 4
Which of these is the best example of Narrow AI?
AA robot that can learn and perform any job a human can do
BA system with human-level reasoning across all domains
CA spam filter that learns to detect junk email from examples
DAn AI that can experience emotions and form lasting memories
Explanation
Narrow AI does one thing well and cannot generalise beyond its training domain. A spam filter is trained on email patterns and excels at that task — but it has zero ability to navigate a car, play chess, or hold a conversation. All AI that exists today is Narrow AI, including GPT-4 and Claude, which are narrow in their own specific ways.
🎉
Phase 1 Complete
4
out of 4
Solid foundation. You understand the fundamentals — the vocabulary and history that everything else builds on.