In everyday language, AI means computer systems that adapt behavior from data or rules to accomplish goals—classification, ranking, translation, planning, or generating content—without hard-coding every branch.
Narrow vs general AI
- Narrow AI — excels at one task (spam filter, face unlock, chess engine)
- General AI (AGI) — hypothetical human-level flexibility across tasks; not productized today
- Product reality — virtually all shipped AI is narrow, often wrapped in UX that feels broader
Rule-based vs learned systems
# Rule-based (explicit logic)
def is_spam_rule(subject: str) -> bool:
banned = ["winner", "free money", "click now"]
return any(w in subject.lower() for w in banned)
# Learned (pattern from examples — conceptual)
# model.fit(labeled_emails) # real training happens elsewhereRules are transparent but brittle; learned models generalize from examples but need data and monitoring.
Practice: Copy optional Python pseudocode into a local .ipynb notebook or REPL. This track is read-focused (execution_profile: none)—no in-browser AI lab; literacy first, not model training.
Where AI appears
Search ranking, fraud detection, medical triage assist, code completion, voice assistants, and content moderation. Each system has inputs, a model or rules engine, outputs, and humans who define success metrics.
Important interview questions and answers
- Q: Is Siri AGI?
A: No—narrow AI orchestrating speech, search, and scripted skills. - Q: AI vs automation?
A: Automation follows fixed scripts; AI adapts from data or learned patterns within a task scope.
Self-check
- Define narrow AI in one sentence.
- Give one example of rule-based vs learned approach.
Tip: When someone says "AI," ask whether they mean rules, classic ML, or Gen AI—clarity prevents mismatched expectations.
Interview prep
- Narrow vs AGI?
- Shipped AI is narrow (one task); AGI is hypothetical general human-level flexibility.
- Rules vs learned?
- Rules are explicit and brittle; learned models generalize from examples but need data and monitoring.