Devices remain learning in incredibly slender strategies, which may result in faults. Deep learning networks need data to solve a certain challenge. If questioned to accomplish a job outside of that scope, it is going to most certainly are unsuccessful. Not enough transparency Learning costs