Human-computer interaction is a combination of several fields. Carnegie Mellon's own Human-Computer Interaction Institute creates technology that improves human capabilities, and it does this by integrating computer science, design, social science and learning science into one discipline. Each field brings something the others don't:
CMU's HCI Institute also notes that its research spans a wide range of topics: innovation in user-interface software tools, studies of computer-supported cooperative work, gesture recognition, data visualization, intelligent agents, etc. That range is evidence for the "multiple disciplines" answer: gesture recognition and human-robot interaction lean heavily on engineering and sensing hardware, while intelligent tutoring systems and cognitive models lean on learning science and psychology.
A weather application featuring a 3D interactive globe animation and smooth gesture controls designed to track hourly precipitation forecasts for the Sahara Desert.
From an interaction design standpoint, this application is an absolute masterpiece. It loads instantly, showcases flawless typography, features intuitive navigation paths and scores a perfect rating across all five of Nielsen's usability components:
However, when evaluated through the lens of utility the app's value collapses entirely. For someone living in a region with virtually zero rainfall, real-time precipitation radars and heavy-downpour alerts solve a problem that simply does not exist. The software achieves peak usability, but because it fulfills no genuine functional need for that specific user, it remains utterly useless.
A university bathroom automatic soap dispenser that is made with no labels, no indicator lights, and a hidden motion sensor. The student has just washed their hands and wants to get a quick pump of soap to finish cleaning up. The main problem is that it breaks Visibility of System Status and gives no clues on how to use it.
Traditional soap dispensers have a pump button you can clearly see and push. But this modern automatic dispenser doesn’t have one. There is no light to show it is on, and no picture or mark to show where the motion sensor is hidden.
When the student puts their hands under the dispenser, nothing happens. They end up waving their hands back and forth in the air, trying to guess where the sensor is or if the soap dispenser is even turned on or out of battery. By adding a small icon right under the dispenser to clearly show where the sensor is and let users know the machine is turned on would make it easier to use.
A prototype is a rough, often incomplete product, built to test an idea before committing the time and cost of building the real thing. Prototypes exist on a spectrum of "fidelity":
The core reason prototypes matter is that testing with real users surfaces problems that the designer cannot predict, because designers already know how their own system is supposed to work and unconsciously fill in gaps that a first-time user cannot.
This is exactly why "Do you like it?" is a weak question. It only captures a polite response (people are often reluctant to criticize something a designer clearly worked hard on), and even when they're honest, "I like it" or "I don't like it" tells you nothing about where someone struggled, hesitated, or misunderstood a label. What "Do you like?" cannot tell a designer: Whether the person could actually complete a real task unassisted, where exactly they got confused or clicked the wrong thing, whether they misunderstood what a button or icon meant, even if they eventually succeeded, etc.
Real usability testing gives people concrete tasks not "what do you think of this app" and watches what actually happens: where they pause or where they say "wait what does this do?". That behavioral data is what a prototype is for.
A smart home voice assistant speaker deployed in a kitchen; Older adults or people with arthritis whose hands shake or lack the fine motor control needed to precisely tap small smartphone screens or press tiny appliance buttons.
Observe users trying to perform everyday kitchen tasks (like setting a cooking timer or checking the weather) using only voice commands from across the room. Measure behavioral metrics like whether they can successfully complete the task without repeating themselves, how often the system misunderstands them and whether they can easily recover when the assistant makes a mistake, all without needing to fall back on a physical screen or keypad.