Let's take an analogy.
Here is a computer, and connected to it by a cable (in this case probably USB, or maybe PS/2) is a keyboard:
[Source: Jeremy Banks, CC BY 2.0, via Wikimedia Commons]
The keyboard is responsible for detecting which key was pressed, and transmitting an agreed/documented "scan code" to the computer. The computer doesn't need to know how the keys are physically arranged.
Here are two more computers, but this time each has a virtual keyboard:
[Source: lolobosse, Mocho (original picture), modification: Mielon, CC0, via Wikimedia Commons]
The role which was previously filled by a separate physical device, is now built into the computer itself, with the functionality it provided implemented in software on top of a generic processor. That software has to convert from the position on the screen the user touched, directly to a character input.
At first glance, the VT-100 in your diagram might look like just a keyboard and a screen, but it is actually performing a more complex job. When the user presses a key, it does not send the VAX-11 scan codes, it translates directly to ASCII characters; and in return, it receives a stream of characters which it must render to the screen, as well as "control codes" which change its output and behaviour.
When you use a "virtual terminal", all of this extra responsibility has moved into software on the computer, just as the responsibility of the physical keyboard moved into the tablet or phone. The software has to convert keyboard activity into ASCII input, and ASCII output into display on the screen, rather than sending and receiving data over a cable and leaving a separate terminal to "understand" it.