This paper presents an artificial neural network (ANN)–based critical review of the ACI 318 Code formula for calculating the effective moment of inertia (Ie) of reinforced concrete (RC) beams. The value of Ie varies along the span length from a gross cross-sectional moment of inertia (Ig) for uncracked sections to a cracked cross-sectional moment of inertia (Icr) for cracked sections. Branson’s expression for Ie was adopted by the ACI 318 Code until 2019 when a new equation for Ie was introduced in the most recent edition of the ACI 318–19 Code; however, the newly proposed modification requires further assessment. For that, the aim of this research is to propose a reasonable modification for estimating Ie based on the ANN technique considering various relevant parameters simultaneously that in turn will lead to more accurate calculation of the deflections of RC beams. Relevant data points were collected from various reputable experimental studies of RC beams and analyzed using the artificial neural network (ANN) technique. The collected data points entailed various parameters with potential influence on the value of Ie including the fibers’ volume fraction (Vf), concrete density, beam depth/width ratio, span length, applied moment (Ma), cracking moment (Mcr), and the values of Icr and Ig. A total of 657 data points were considered in the ANN training, testing, and verification processes. The results showed that the value of Ie noticeably increases with the increase in the Vf, concrete density, depth/width ratio, and Icr. Additionally, the influence of Mcr/Ma was more pronounced at a ratio of 0.5 rather than 2/3 as specified in the ACI-318–19 Code, which is included in the proposed modification to the ACI-318–19 Code.