← Back to course
AI and You 9-12 / Week 07 / Wednesday
3/6
Week 07 · Privacy in the Machine

Wednesday

Explorer Lab: Read the log
// What a question log reveals
⏱ about 20 min

Wednesday: Explorer Lab: Read the Log

The builder sends a sample of the proposed log: twelve rows from one week. The packet's copy had names.

Librarian Joss replaced each name with an ID before the crew saw it. "No names leave my desk," she says.

Comet grins. "Great, so it is anonymous now. Nothing to worry about!"

Wren is already counting. "Maybe. What do you notice about the times?"

Comet scans the Time column. "One student asks a lot of questions really late."

Nova projects the twelve rows onto the wall. "Would you like a hint?" she asks. "Pretend you also had the class lists. Would the ID still hide anyone?"

Wren hands you a pencil. "Let's find out what this log really gives away."

Your mission

Tally the log, look for patterns, and test whether an ID really hides a person. Then rebuild the log so it keeps only what SH-1 needs.

LOG-12 is made-up data from the SH-1 story. Students are S01 to S40, never named.

  • Paper and a pencil
  • The LOG-12 table below
  • The two class lists below
RowStudentClassDayTimeTopic
L01S03BiologyMon4:10 pmcell parts
L02S07SpanishMon11:40 pmverb endings
L03S11AlgebraMon7:05 pmslope
L04S03BiologyTue4:15 pmcell parts
L05S07SpanishTue11:35 pmverb endings
L06S02World HistoryTue6:30 pmtrade routes
L07S11AlgebraWed7:00 pmslope
L08S07SpanishWed11:45 pmverb endings
L09S05BiologyWed3:50 pmphotosynthesis
L10S02World HistoryThu6:20 pmtrade routes
L11S07AlgebraThu11:50 pmslope
L12S11AlgebraThu7:10 pmslope
PREDICT FIRST
  • Read the question.
  • Tap your answer.
Before you tally: which topic do you think appears most often?
  1. On paper, write the five topics in a column.
  2. Go down LOG-12 row by row and make a tally mark next to each row's topic.
  3. Add the tallies and check that they total 12.
  4. Now list each student ID and the times they asked.
  5. Circle every row asked after 11:30 pm.
  6. Look at the class lists below and find who takes both Spanish and Algebra.
  7. Finally, write a new table with only the topics and their totals.
Class list (made up)Students
Spanish, Teacher MaraS04, S07, S09, S15, S21
Algebra, Teacher DevS07, S11, S12, S18, S25
CHECK YOUR TALLIES
  • Read the question.
  • Tap your answer.
How many slope questions are in LOG-12?
How many verb endings questions are there?
Add all five topic totals: 4 plus 3 plus 2 plus 2 plus 1. What do you get?
FIND THE PATTERN
  • Read the question.
  • Tap your answer.
How many rows were asked after 11:30 pm?
Which student asked every one of those late rows?
S07 asked about Spanish and Algebra. Using the class lists, who takes both?

What the lab shows

NIST warns that AI systems can create new privacy risks by inferring who someone is, or private facts about them.

NIST also says models may stitch together pieces of information to guess private facts.

Here, a late-night pattern plus a class list points to one person, even with no name in the log.

Lab findingTrue or false?
Replacing names with IDs made it impossible to tell who S07 is.?
The topic totals still show that slope is the most asked topic.?
The topic totals table shows who studies late at night.?
Combining two data sets can reveal more than either one alone.?
WHY THIS EXERCISETotals keep the useful pattern and drop the details that point to a person.
On paper, draw your topics-only table: five topics and their totals, with no IDs, days or times.

Excellent lab work, reviewer. Tomorrow you will weigh each privacy fix and what it gives up.

← Tuesday