The exact text we saved

This is the version exactly as it was saved, character for character. The SHA-256 hash below is a 64-character code worked out from that text. Run the same text through SHA-256 and you get the same code; change a single letter and the code comes out completely different. Each version also stores the previous version’s hash, which chains them together so no version can be swapped out quietly.

TRUVACE RECORD VERSION
record: TRV-2026-2798
version: 1
kind: certified
reason: Certified into the record
timestamp: 2026-11-12T17:25:18.843067Z
status: published
lens: g_space
sector: education
headline: Effects of AI feedback on students’ English writing performance in higher education: a meta-analysis
dek: The integration of artificial intelligence (AI) to assess English writing has become prevalent in higher education, though empirical research on its efficacy has yielded inconsistent results. To clarify this variability and identify moderating factors, this study conducted a three-level meta-analysis synthesizing 234 effect sizes from 43 valid studies. The analysis demonstrated a significant, moderate overall impact of AI feedback on writing performance (g = 0.58, p < 0.001). The finding also revealed that AI fe…
gain_title: A meta-analysis of 43 studies (234 effect sizes) found that AI feedback raised university students' English writing performance by a moderate margin (g = 0.58), with larger gains when paired with teacher feedback.
problem_title: (none)
trace_subject: (none)
gain_reading: A meta-analysis of 43 studies (234 effect sizes) found that AI feedback raised university students' English writing performance by a moderate margin (g = 0.58), with larger gains when paired with teacher feedback.
gain_evidence: significant, moderate overall impact of AI feedback on writing performance (g = 0.58, p < 0.001)
problem_reading: (none)
problem_evidence: (none)
quick_read: A three-level meta-analysis pooled 234 effect sizes from 43 studies on AI-based assessment of English writing in higher education. It found a significant, moderate overall benefit for writing performance (g = 0.58, p < 0.001).

Earlier research on AI writing feedback had produced inconsistent results, which the analysis set out to explain by testing moderating factors. Feedback worked better when combined with human input, especially from teachers, and the biggest effect appeared on narrative writing tasks.
limitation: 
tag: Evidence-backed gain
key_points: The analysis combined 234 effect sizes from 43 valid studies using a three-level meta-analytic model. | The overall effect of AI feedback on English writing performance was moderate and significant (g = 0.58, p < 0.001). | Feedback was more effective when AI was combined with human feedback in a hybrid intelligence framework. | The pairing of AI with teacher feedback was singled out as particularly effective. | AI had its largest effect on narrative writing tasks.
rundown: The study addressed inconsistent results in earlier research on AI tools that assess English writing in universities. By pooling 234 effect sizes from 43 studies in a three-level model, it estimated an overall effect of g = 0.58, described as moderate and statistically significant.

Moderator analysis showed that outcomes depended on how feedback was delivered. Hybrid arrangements that combined AI with human feedback, notably teacher feedback, outperformed AI alone, and narrative writing tasks showed the largest gains from AI feedback.
sources:
- peer_reviewed | Cogent Education | https://doi.org/10.1080/2331186x.2026.2665494 | 2026-04-30
prev: 0000000000000000000000000000000000000000000000000000000000000000

SHA-256 hash

d257f99b1baf764022032634ca95c2ffeb916ea707e42532e808080f776a854e

Previous version’s hash

0000000000000000000000000000000000000000000000000000000000000000