1[2 {"id":"C1","text":"Of the five studies, five printed values were extracted, four could be harmonised to theta = ln(T_AI/T_noAI), and two (S1, S4) could be pooled; S5 measures output counts in a DiD and is not harmonisable.","results_keys":["k_extracted","k_harmonised","k_pooled","items"],"status":"supported","caveat":"Poolability limited by what the extracts print; S2 and S3 uncertainty exists in the full papers but is not admissible here."},3 {"id":"C2","text":"Harmonised estimates: S1 -0.240, S2 +0.174, S3 -0.354, S4 -0.118 (log time ratio).","results_keys":["items"],"status":"supported","caveat":"S3 and S4 use the printed percentages, which differ slightly from ratios of displayed means."},4 {"id":"C3","text":"The pooled random-effects estimate over S1 and S4 is theta = -0.188 (about 17% shorter completion time), 95% CI [-0.396, 0.020], equal to the fixed-effect estimate.","results_keys":["re_estimate","re_se","re_ci_low","re_ci_high","fe_estimate","fe_ci_low","fe_ci_high","pooled_pct_change_re"],"status":"supported","caveat":"k = 2; S4 SE approximated from a p-value."},5 {"id":"C4","text":"The pooled effect is not statistically significant at the 5% level; the evidence does not establish that AI assistance shortens task time.","results_keys":["pooled_significant_5pct","re_ci_low","re_ci_high"],"status":"supported","caveat":"Absence of significance with k=2 is not evidence of no effect."},6 {"id":"C5","text":"Measured heterogeneity among the pooled studies is low (Q = 0.324, I2 = 0%, tau2 = 0), but this reflects k = 2 and the excluded studies' estimates (+0.174, -0.354) suggest substantive heterogeneity not captured by the pooled analysis.","results_keys":["q_stat","i2","tau2","heterogeneity_interpretation","items"],"status":"supported","caveat":"Heterogeneity of excluded studies is qualitative only; cannot be quantified without SEs."},7 {"id":"C6","text":"The pooled result is robust to using the CI-derived SE for S1 and the displayed-means ratio for S4.","results_keys":["sensitivity_ci_se_s1_re_estimate","sensitivity_ci_se_s1_re_se","sensitivity_ratio_s4_re_estimate","sensitivity_ratio_s4_re_se","sensitivity_ratio_s4_theta"],"status":"supported","caveat":"Only two small perturbations examined."},8 {"id":"C7","text":"Baseline unweighted means of theta are -0.179 (pooled studies) and -0.134 (all harmonised studies).","results_keys":["baseline_unweighted_mean_theta_pooled","baseline_unweighted_mean_theta_harmonised"],"status":"supported","caveat":"Unweighted means ignore precision; descriptive only."},9 {"id":"C8","text":"The project is complete in the TASK sense: every item has harmonised values or a justified statement of what cannot be obtained.","results_keys":["feasibility","feasibility_reason","data_issues"],"status":"supported","caveat":"Completeness of the project does not imply strength of evidence."}10]11