The Role of Statistics in the Advancement of Artificial Intelligence and Modern Technology

<title>Abstract</title> <p>This study provides an empirical assessment of the historical evolution and projected future trends of Statistics and Statistical Methods in Modern Technology and Artificial Intelligence (AI), with a focus on their interdependence. Using publication data extracted from Semantic Scholar (1991–2023), grouped based on publication data on Statistics in Modern Technology and Statistical Methods in AI, the analysis adopts a two-stage methodological approach. First, a machine learning–driven time series model is developed using the Random Forest algorithm. The series was rigorously tested for stationarity using Augmented Dickey-Fuller (ADF) and KPSS tests, revealing that Statistical Methods in AI exhibits trend-stationarity around a deterministic linear trend (R² = 0.949). After detrending, the series follows a clear AR(1) process (PACF significant only at lag 1, φ₁ = 0.760). Feature engineering incorporated autoregressive lags and trend components, with time series cross-validation (expanding window) used for robust evaluation. Hyperparameter tuning optimized model performance (n_estimators = 264, max_depth = None), and conformal prediction generated 90% uncertainty intervals. Second, Spearman's rank correlation analysis examines the relationship between the two series. The model demonstrates strong performance (training MAE = 0.0736 on detrended scale) with lag-1 dominating feature importance (75.4%). A structural break was identified during 2015–2017, where prediction errors peaked at 0.4656. Forecasts indicate continued exponential growth, with Statistical Methods in AI projected to reach 2,682 publications by 2030 (90% CI: 2,386–3,015). For Statistics in Modern Technology, previous analysis projects values between 427 and 786 within the same period. Spearman's rank correlation analysis reveals an exceptionally strong, positive, and statistically significant monotonic association between the two variables (ρ = 0.985, p &lt; 0.001), demonstrating their tightly coupled progression. The findings affirm that statistics functions not merely as a complementary tool but as a central driver of technological and AI advancement. The AR(1) structure of the detrended series suggests that deviations from the long-term trend correct slowly (half-life ≈ 2.5 years), with 76% of each year's deviation persisting to the next. The structural break during 2015–2017 warrants further investigation into potential external factors (e.g., shifts in AI research focus, publication patterns). The study recommends strengthened interdisciplinary collaboration and increased investment in statistical innovation to support sustainable and responsible technological development.</p>