I like the general advice of learning by using only manual mode. However, automatic mode wasn't the cause of any problems with this particular picture.

The basic problem is that this scene has a very wide dynamic range, which is pretty much true any time the light sources illuminating the scene also appear in the picture. This is no different from the sun appearing in a normal daytime outdoor picture. To keep the light sources from being overexposed and blowing out, the rest of the scene must by necessity be underexposed. There are basically two ways to deal with high dynamic range: use a sensor with high dynamic range or take multiple pictures at different exposures, then do intelligent blending afterwards. The latter is often referred to as "HDR" (high dynamic range) processing.

Modern cameras do capture impressively high dynamic range compared to film, and are getting steadily better. However, to access that, you have to capture in RAW mode and do all the manipulations in that high range space before converting to the low range space the picture will ultimately be displayed in.

Even your JPG with its limited dynamic range contains a lot more details in the dim areas than are apparent in your original picture. Here is what I got after some fairly aggressive non-linear brightening:

Note the splotches on the pavement in the foreground. These are because the original didn't have more resolution than this in these dim areas. The sudden transition between two shades of gray are because these are differences of one count out of 8 bits in your original. This small difference in your original was greatly amplified by the brightening I applied to the picture after the fact. If you go back to your original 12 or 14 bit raw image, you can apply the same brightening without the result being splotchy like this. If you only kept the JPG you showed us, then there is nothing more you can do because the extra information your sensor captured has already been discarded.

This picture is a great example of why to always shoot RAW, then post-process in that space, then convert to the limited 8 bits/color final display space as the last step only. Think of it this way. If your camera captures 14 bit values, then there are 214 = 16000 intensity levels per pixel. By converting this to 8 bits (256 levels), you are picking one level from 64 on average. 256 levels per color per pixel is more than most mediums can display, so is good enough as the final image in most cases. However, the higher input resolution of RAW allows you to chose those 256 levels non-linearly from from the original with smooth input steps corresponding to each output step. The lack of that is exactly why you see splotches in my brightened version of your picture.