Imagine a graph vs just the raw data. Which one is easier to pull the information from and comprehend? Our brains are purpose built to rapidly digest and comprehend a constant influx of visual data and ways to convert raw output into easy to comprehend visualizations dramatically increase a general user's ability to interface with a system and the data that it generates. The dashboard of your car is a GUI, for example.

Also, there's the humble button. A single graphic combined with a single mouse button that can be used to perform a near infinite number of actions. It's a translation layer that allows you to perform actions preset by the company who designed the thing so that a user doesn't have to already be familiar with how it runs in order to accomplish everyday tasks. I could type "@self -cast 'light attack' -target 'Malenia, Blade of Miquella'" over and over again, but why would I when I can just spam click the right shoulder button on my controller until I die a very quick death?

The issue that Linux users run into (and why Linux has never reached any real traction in the global computer user base) is that they're usually power users who already use a terminal or are well versed in similar systems. They don't necessarily need a way to translate data to or from the terminal because they've been using it long enough that it feels simple, like a physicist wondering why cars measure the fuel tank with a dial gauge rather than a number showing the newtons of potential energy left in the gas tank. It's a meaningless number to the average person, who just wants to know whether or not they have gas in the tank and maybe the distance that will let them travel before the tank runs out. You could calculate the latter using the potential energy in the tank, but the average person isn't going to know that formula and be able to bust it out at any given time, so there's a GUI in the car that does it for you.