ai-engineering-from-scratch/phases/19-capstone-projects/17-personal-ai-tutor/quiz.json at main · Amjad-AbuRmileh/ai-engineering-from-scratch · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
{
  "lesson": "17-personal-ai-tutor",
  "title": "Capstone 17 — Personal AI Tutor (Adaptive, Multimodal, with Memory)",
  "questions": [
    {
      "stage": "pre",
      "question": "What does a Socratic tutor policy do when a learner asks for the answer outright?",
      "options": [
        "Asks a leading question that scaffolds the learner toward the answer",
        "Logs the request and goes silent",
        "Provides the answer immediately and moves on",
        "Switches to a different concept node"
      ],
      "correct": 0,
      "explanation": ""
    },
    {
      "stage": "pre",
      "question": "Which structure does the curriculum graph encode?",
      "options": [
        "A directed graph of concepts with prerequisite edges, attached to OER content",
        "A random shuffle of topics",
        "A flat list of unrelated concepts",
        "Only the most recently studied node"
      ],
      "correct": 0,
      "explanation": ""
    },
    {
      "stage": "check",
      "question": "What does the learner model update after each interaction?",
      "options": [
        "The base LLM's weights",
        "The COPPA retention policy",
        "The curriculum graph schema",
        "Per-concept mastery probability via Bayesian knowledge tracing or a similar variant"
      ],
      "correct": 3,
      "explanation": ""
    },
    {
      "stage": "check",
      "question": "How is memory split between episodic and semantic stores?",
      "options": [
        "Episodic holds every interaction; semantic holds compacted mistakes and preferences promoted from episodic",
        "They are duplicates of the same data",
        "Episodic is for the parent; semantic is for the learner",
        "Episodic holds embeddings; semantic holds raw audio"
      ],
      "correct": 0,
      "explanation": ""
    },
    {
      "stage": "check",
      "question": "Which constraint shapes memory retention for K-12 learners?",
      "options": [
        "Retention based on model context length",
        "Permanent retention with no deletion",
        "GDPR only",
        "COPPA-aware retention with auto-deletion after a defined window and parental access"
      ],
      "correct": 3,
      "explanation": ""
    },
    {
      "stage": "post",
      "question": "What is the structure of the efficacy study used to gate the deliverable?",
      "options": [
        "Anonymous internet survey",
        "A 30-second AB test",
        "Pre-test and post-test over two weeks with 10 learners, compared against a non-adaptive baseline cohort",
        "A single one-hour session with one learner"
      ],
      "correct": 2,
      "explanation": ""
    },
    {
      "stage": "post",
      "question": "Which multimodal input paths does the tutor expose to the learner?",
      "options": [
        "Text only",
        "Photo only via a custom OCR model",
        "Voice only with no text or photo",
        "Text typed, voice via LiveKit + Whisper, and photo math via dots.ocr or PaliGemma 2"
      ],
      "correct": 3,
      "explanation": ""
    }
  ]
}