Count something real: a 3-day field observation study
Before dashboards, data science was a person with a clipboard. Physically count something real — canteen queues, vehicles at a crossing, library seats — at fixed intervals over 3+ days, then put AI on the numbers to find the pattern.
Posted by The AI Internships
The work brief
- 01Pick something countable you can observe repeatedly, and design the protocol: what exactly counts, at which times, from where.
- 02Observe at 3+ fixed time slots per day for 3+ days. Record on the spot, not from memory. Photograph your tally sheet or location once.
- 03Enter the data as a CSV, then ask AI to hunt for patterns: peak times, day differences, anomalies. Verify its claims against your tally sheet — you were there; it wasn’t.
- 04Answer one useful question with it: when should someone come to avoid the queue?
What you’ll produce
Submission standard
Submit the protocol, your observations as CSV, a photo from your observation spot, and the pattern findings. Include your AI workflow — tools, best prompts, and what you changed. Your photo must show your dated, handwritten tally sheet (not just the location) with marks that add up to your CSV counts.
The protocol: what counts, observation times, your vantage point
Written responseRequiredYour observations as CSV: date, time slot, count, notes
date, time slot, count, notes — record on the spot; note anomalies and their real-world causes.
CSV dataRequiredPhoto of your tally sheet or observation spot
Photograph your handwritten tally sheet showing dated, on-the-spot marks (not just the location) — the marks should add up to your CSV counts.
Image uploadRequiredThe patterns, the anomalies with real-world explanations, and your useful answer
Written responseRequired
You’ll complete these inside your private workspace.
How your work is evaluated
The passing benchmark is 70/100.
Real fieldwork
29%The data shows real observation — natural variance, anomalies with explanations.
Protocol quality
14%Consistent definitions and times make the data comparable.
Findings
29%Patterns are numeric and the useful answer follows from them.
Dated tally ↔ CSV consistency
29%The photographed tally sheet must show dated, on-the-spot marks that reconcile with the CSV counts, and the findings must follow from those rows. Suspiciously smooth counts with no anomalies and an undated/location-only photo are a fail.
How we grade your AI usage
Using AI is the point — it’s the skill this certificate proves. You’ll answer three short questions about how you used it: what you asked, what was wrong with its first answer, and what you changed. Specific, honest answers score high. “I pasted the brief and submitted the answer” scores near zero.