<p>Multitasking optimization (MTO) has recently received extensive attention due to its closer resemblance to the multiple-task scenarios encountered in real-world settings. Extracting transferable knowledge between tasks is the key to solving MTO problems, as it accelerates the resolution of each individual task. Most research considers task interactions to be symmetrical and employs calculated similarity to determine the degree of interaction. However, task interactions are frequently asymmetrical in real-world scenarios, one task may have a greater or lesser impact on another task. So, an online asymmetric knowledge transfer evolutionary algorithm is proposed for solving MTO problems. This algorithm employs an asymmetric knowledge transfer matrix and updates it at regular intervals. We evaluate the performance of our proposed algorithm against several state-of-the-art algorithms on multi-objective multitasking test problems and multi-level inverter cases. The experimental results reveal that the proposed algorithm demonstrates strong competitiveness in solving MTO problems.</p>