We present a low complexity algorithm for detection in large MIMO systems based on Hopfield neural network (HNN) algorithm. Our algorithm achieves near Maximum Likelihood performance for high number of antennas. It has much lesser computational and algorithmic complexity compared to ML decoding and lesser computations for each searching step compared to conventional Likelihood Ascent Search (LAS) algorithm. No conditional evolution, comparison to multiple thresholds and multi-level quantizations are necessary for updating in the search stage. We present simulations confirming that, with simpler evolution and lesser computations, our algorithm yields comparable results to LAS algorithm.