Skip to content

R0031/2026-03-29/C004 — Assessment

BLUF

The 22% figure is confirmed from the BestColleges March 2023 survey. However, the figure is rapidly dated — the same survey series found 56% by October 2023, and other surveys report even higher rates.

Probability

Rating: Very likely (80-95%)

Confidence in assessment: Medium

Confidence rationale: The 22% figure is confirmed but the claim lacks source attribution, which introduces uncertainty about whether the intended source is BestColleges or another survey with a similar finding.

Reasoning Chain

  1. FACT: BestColleges surveyed 1,000 US college students in March 2023. Of all respondents, 43% had used AI tools, and 50% of those used them on assignments, yielding 22% of the total sample. [SRC01-E01, Medium reliability, High relevance]

  2. REPORTED: By October 2023, BestColleges found 56% of college students had used AI on assignments or exams. [SRC01-E01, Medium reliability, High relevance]

  3. JUDGMENT: The 22% figure was accurate at the time of the March 2023 survey but represents a rapidly moving target.

Evidence Base Summary

Source Description Reliability Relevance Key Finding
SRC01 BestColleges March 2023 survey Medium High 22% of students used AI on assignments

Collection Synthesis

Dimension Assessment
Evidence quality Medium — single survey, moderate sample size
Source agreement N/A — single source identified for the exact figure
Source independence N/A
Outliers Later surveys show much higher rates

Gaps

Missing Evidence Impact on Assessment
No source attribution in the original claim Medium — cannot confirm intended source
Single survey methodology Medium — 1,000 respondents, US-only

Researcher Bias Check

Declared biases: The 22% figure supports the article's narrative about hidden AI use.

Influence assessment: Low risk for the specific number, but the researcher should note that this figure is likely outdated.

Cross-References

Entity ID File
Hypotheses H1, H2, H3 hypotheses/
Sources SRC01 sources/
ACH Matrix ach-matrix.md
Self-Audit self-audit.md