TIG welding with activating flux, known as Activated Tungsten Inert Gas (ATIG) welding, is used to overcome low penetration and to increase productivity in contrast to conventional TIG welding. In this work, 20 experimental runs were done with autogenous butt welding on 304SS base plate with a thickness of 5 mm. The Response Surface Methodology (RSM) with Central Composite Design (CCD) was taken into consideration to determine the treatments for carrying out experiment. A binary flux mixture comprising MoO3 and MnO2 mixed in the ratio of 4:1, 2.5:1, and 1:1 was considered for conducting the experiment. The maximum penetration of 4.703 mm was attained using a 0.577 kJ/mm heat input, a gas flow rate of 6 l/min, and a flux ratio of 4:1, as per the findings. This investigation employs an Artificial Neural Network (ANN) approach to estimate ATIG weldment penetration. Application of ANN in ATIG welding was not yet reported in the literature. The binary flux mixture comprising molybdenum oxide and manganese dioxide mixed in different ratios is also newly explored in the present investigation. These are the novelty of the present work. Here, process characteristics including heat input, flux ratio, and gas flow rate are included in the input layers of the ANN model used for training, while bead width and penetration are taken into account in the output layer. From the present investigation, it can be stated that the ANN predicted values are consistent with that of the experimental data.