This paper investigates user scheduling in a slotted downlink multi-user multiple-input multi-output (MU-MIMO) system, where a base station (BS) equipped with multiple antennas transmits status updates to multiple single-antenna devices. We assume perfect channel station information (CSI) is available at the BS to make scheduling decisions. The goal is to optimize the information freshness of the network, quantified by the network-wide average age of information (AoI), subject to the peak transmission power constraint in each time slot. To tackle this problem, we employ the Lyapunov optimization method to convert the constrained long-term optimization problem into a per-slot optimization problem. We propose two low-complexity algorithms to solve the per-slot optimization problem, namely the AoI-aware scheduling (AS) algorithm and the joint age- and channel-aware scheduling (JACS) algorithm. These algorithms aim to determine the optimal device or devices for delivering the latest status updates to them in each time slot. We evaluate the performance of the proposed algorithms through numerical simulations in both symmetric and asymmetric networks and compare them with two baseline algorithms: semi-orthogonal user group (SUG) and exhaustive search (ES) algorithms. Regardless of the network setup, our proposed algorithms consistently outperform SUG, and demonstrate performance comparable to that of the exhaustive search algorithm. Notably, when the peak transmission power is high enough, the proposed algorithms perform almost as well as the ES algorithm. Thanks to the joint consideration of CSI and AoI, JACS tends to have better performance than AS at the cost of slightly higher computational complexity.