WEBVTT 00:00:00.000 --> 00:00:05.379 So picture this. It's 1979, and the alarms at 00:00:05.379 --> 00:00:08.000 the Three Mile Island nuclear plant are just, 00:00:08.160 --> 00:00:10.220 well, they're screaming. Right, an absolute nightmare 00:00:10.220 --> 00:00:12.720 scenario. Yeah, completely. But the operators 00:00:12.720 --> 00:00:14.439 standing in the control room, they actually didn't 00:00:14.439 --> 00:00:15.900 know what the system was trying to say to them. 00:00:16.100 --> 00:00:19.359 Exactly. And it wasn't just some mechanical failure. 00:00:19.480 --> 00:00:22.800 It was a deadly failure of language. Between 00:00:22.800 --> 00:00:26.140 human and machine. Precisely. Yeah. That communication 00:00:26.140 --> 00:00:28.699 breakdown is basically a textbook example of 00:00:28.699 --> 00:00:32.659 why human -computer interaction, or HCI, is about 00:00:32.659 --> 00:00:35.719 so much more than like keyboards and mice and 00:00:35.719 --> 00:00:38.820 glowing screens. Investigations into that disaster 00:00:38.820 --> 00:00:40.780 actually concluded that the design of the human 00:00:40.780 --> 00:00:43.359 -machine interface was partly responsible for 00:00:43.359 --> 00:00:45.880 the meltdown itself. Which is wild to think about. 00:00:46.000 --> 00:00:48.000 Right. The system was generating all this critical 00:00:48.000 --> 00:00:50.880 data, but the interface, it just failed to translate 00:00:50.880 --> 00:00:53.140 that data into a format that the humans could 00:00:53.140 --> 00:00:56.280 rapidly parse during a massive crisis. It fundamentally 00:00:56.280 --> 00:00:58.939 failed them. It did. The interface fundamentally 00:00:58.939 --> 00:01:01.340 failed the operators. OK, let's unpack this. 00:01:02.159 --> 00:01:04.040 Because framing our relationship with machines 00:01:04.040 --> 00:01:07.219 as a literal language like an ongoing dialogue 00:01:07.219 --> 00:01:11.409 that completely shifts. how you look at the technology 00:01:11.409 --> 00:01:13.230 you use every day. It really does change your 00:01:13.230 --> 00:01:16.150 perspective. It does. So welcome, everyone, to 00:01:16.150 --> 00:01:19.109 today's Deep Dive. We are exploring the hidden 00:01:19.109 --> 00:01:21.950 architecture of how we actually communicate with 00:01:21.950 --> 00:01:25.469 the digital world. And as a fellow curious learner, 00:01:25.650 --> 00:01:27.769 I just think it's fascinating how formalized 00:01:27.769 --> 00:01:29.769 this study has become. Oh, absolutely. We are 00:01:29.769 --> 00:01:32.670 looking at a discipline that treats using a computer 00:01:32.670 --> 00:01:35.269 not like swinging a hammer, but like having an 00:01:35.269 --> 00:01:38.069 open -ended conversation. Yeah, and the terminology 00:01:38.069 --> 00:01:40.459 itself really reflects that shift in thinking. 00:01:41.099 --> 00:01:43.159 The first known use of the term human -computer 00:01:43.159 --> 00:01:45.819 interaction was by a researcher named Carlisle 00:01:45.819 --> 00:01:49.319 back in 1975. Okay, so mid -70s. Right. But it 00:01:49.319 --> 00:01:51.180 truly gained traction and became a recognized 00:01:51.180 --> 00:01:54.340 discipline after the 1983 publication of this 00:01:54.340 --> 00:01:57.159 seminal book by Stuart Card, Alan Newell, and 00:01:57.159 --> 00:01:59.269 Thomas P. Moran. What did they figure out that 00:01:59.269 --> 00:02:01.189 was so groundbreaking? Well, they established 00:02:01.189 --> 00:02:03.750 the foundational concept that a computer has 00:02:03.750 --> 00:02:07.569 such a vast, malleable range of uses that operating 00:02:07.569 --> 00:02:10.310 one is best understood as a dynamic dialogue. 00:02:10.569 --> 00:02:13.469 A dialogue. Right. You input a command, the system 00:02:13.469 --> 00:02:16.509 interprets it, alters its state, and then communicates 00:02:16.509 --> 00:02:18.710 the new state back to you. And when that loop 00:02:18.710 --> 00:02:21.009 breaks? Yeah, when that loop is broken or when 00:02:21.009 --> 00:02:24.219 the translation fails. The results scale from, 00:02:24.219 --> 00:02:27.580 you know, mild frustration to absolute catastrophe. 00:02:27.860 --> 00:02:30.139 Which brings us right back to the physical stakes 00:02:30.139 --> 00:02:32.659 of a bad interface. We saw it with Three Mile 00:02:32.659 --> 00:02:35.180 Island, and there are honestly chilling examples 00:02:35.180 --> 00:02:37.819 of this in aviation as well. Oh, aviation is 00:02:37.819 --> 00:02:40.199 a major one. Right. Like manufacturers will occasionally 00:02:40.199 --> 00:02:42.680 update flight instruments or change the physical 00:02:42.680 --> 00:02:45.139 layout of the throttle quadrants and cockpits. 00:02:45.780 --> 00:02:48.060 And conceptually on paper, these new designs 00:02:48.060 --> 00:02:50.319 are engineered to be mathematically superior. 00:02:50.740 --> 00:02:52.580 They are theoretically optimized. for better 00:02:52.580 --> 00:02:54.960 human -machine interaction. But then reality 00:02:54.960 --> 00:02:57.180 hits. Exactly. When they are rolled out, they 00:02:57.180 --> 00:03:00.349 end up causing fatal accidents. Why does an objectively 00:03:00.349 --> 00:03:02.689 quote unquote better design cause a plane to 00:03:02.689 --> 00:03:05.050 crash? It really comes down to cognitive load 00:03:05.050 --> 00:03:09.009 and ingrained heuristics. OK, break that down 00:03:09.009 --> 00:03:11.530 for me. Well, the human side of the interaction 00:03:11.530 --> 00:03:13.590 equation has already adapted to the previous 00:03:13.590 --> 00:03:17.150 interface. Pilots spend thousands of hours building 00:03:17.150 --> 00:03:20.210 physical and mental models of a specific cockpit 00:03:20.210 --> 00:03:24.289 layout. Muscle memory. Exactly. In high stress, 00:03:24.689 --> 00:03:26.849 split -second environments like recovering from 00:03:26.849 --> 00:03:29.909 an aerodynamic stall, there's literally no time 00:03:29.909 --> 00:03:33.550 for conscious, deliberate analysis of a new dashboard. 00:03:33.840 --> 00:03:37.340 You just react? Right. You rely entirely on ingrained 00:03:37.340 --> 00:03:39.780 behaviors. So when the interface has secretly 00:03:39.780 --> 00:03:42.719 changed the rules, the pilot's automatic response 00:03:42.719 --> 00:03:45.460 misaligns with the machine's new input requirements. 00:03:45.800 --> 00:03:47.500 It makes total sense when you think about it 00:03:47.500 --> 00:03:49.400 in terms of pure muscle memory. I mean, it's 00:03:49.400 --> 00:03:51.719 like renting a car, driving down the highway, 00:03:51.900 --> 00:03:54.000 and suddenly realizing the manufacturer decided 00:03:54.000 --> 00:03:56.300 to swap the brake and the gas pedals. That is 00:03:56.300 --> 00:03:58.280 a terrifying thought. But yeah, it's a perfect 00:03:58.280 --> 00:04:00.500 analogy. It honestly doesn't matter if the new 00:04:00.500 --> 00:04:02.680 pedals are made of aerospace -grade materials, 00:04:02.680 --> 00:04:06.539 or if their placement is anatomically more ergonomic 00:04:06.539 --> 00:04:09.080 for your ankle. Your foot is going to stomp where 00:04:09.080 --> 00:04:11.580 it expects the brake to be. Right. And you are 00:04:11.580 --> 00:04:14.099 going to crash. Yeah. If we connect this to the 00:04:14.099 --> 00:04:17.100 bigger picture, it illustrates why HCI is fundamentally 00:04:17.100 --> 00:04:19.879 about ensuring humans can safely and efficiently 00:04:19.879 --> 00:04:23.300 interact with incredibly complex technologies. 00:04:23.399 --> 00:04:26.019 It's way beyond making things look pretty. Exactly. 00:04:26.220 --> 00:04:29.180 It is not about aesthetics. In high -stakes fields 00:04:29.180 --> 00:04:31.920 like aviation, healthcare, industrial control, 00:04:32.439 --> 00:04:35.360 the interface is the single point of translation 00:04:35.360 --> 00:04:38.540 between human intent and mechanical action. So 00:04:38.540 --> 00:04:41.360 if a misunderstood dialogue causes disasters, 00:04:41.720 --> 00:04:44.660 how do engineers ensure this loop of interaction 00:04:44.660 --> 00:04:47.100 actually flows correctly? That is the million 00:04:47.100 --> 00:04:49.560 dollar question. Right. Because information is 00:04:49.560 --> 00:04:51.459 supposed to flow from the human to the computer. 00:04:51.660 --> 00:04:54.220 The computer evaluates it, processes it, and 00:04:54.220 --> 00:04:57.180 then sends feedback back to the human. But the 00:04:57.180 --> 00:04:59.139 channels being used to accomplish this are becoming 00:04:59.139 --> 00:05:01.639 incredibly multi -dimensional now. Oh, yeah. 00:05:02.279 --> 00:05:03.980 We aren't just typing commands into a terminal 00:05:03.980 --> 00:05:06.699 anymore. The interaction loop now relies on a 00:05:06.699 --> 00:05:09.160 really complex web of simultaneous channels. 00:05:09.279 --> 00:05:11.800 Like what? Well, visual -based interaction is 00:05:11.800 --> 00:05:14.620 the most widespread, but it has evolved far beyond 00:05:14.620 --> 00:05:16.439 just clicking a mouse. Right. We have touch screens 00:05:16.439 --> 00:05:18.600 and all that. And even more advanced than that, 00:05:19.079 --> 00:05:21.480 it encompasses large -scale body movement tracking. 00:05:21.709 --> 00:05:24.649 facial expression analysis to gauge user intent, 00:05:25.490 --> 00:05:28.709 and gaze detection. Gaze detection, like it watches 00:05:28.709 --> 00:05:31.649 where I look. Exactly. The system is literally 00:05:31.649 --> 00:05:34.269 tracking the micro movements of your eyes to 00:05:34.269 --> 00:05:36.730 anticipate what information you need before you 00:05:36.730 --> 00:05:39.410 even request it. Wow. And there is a massive 00:05:39.410 --> 00:05:41.860 audio component as well, which... I mean, it 00:05:41.860 --> 00:05:44.060 isn't just basic voice commands anymore. Right. 00:05:44.100 --> 00:05:46.939 Audio -based interaction is acquiring and processing 00:05:46.939 --> 00:05:50.100 complex audio signals. Yes. But what's fascinating 00:05:50.100 --> 00:05:52.519 is how these channels now cross -reference each 00:05:52.519 --> 00:05:55.300 other to ensure accuracy. Wait, how do they cross 00:05:55.300 --> 00:05:57.480 -reference? For instance, visual approaches, 00:05:57.680 --> 00:06:00.540 like tracking lip movements, are highly influential 00:06:00.540 --> 00:06:02.800 in correcting errors in audio speech recognition. 00:06:02.920 --> 00:06:05.600 You're kidding. Nope. The computer is essentially 00:06:05.600 --> 00:06:07.980 watching the physical shape of your mouth to 00:06:07.980 --> 00:06:10.459 verify that it correctly parsed the audio frequencies 00:06:10.459 --> 00:06:13.430 of your That is incredible. The machine is cross 00:06:13.430 --> 00:06:16.149 -referencing my vocal cords with my facial musculature 00:06:16.149 --> 00:06:18.350 to make sure we aren't having a miscommunication. 00:06:18.790 --> 00:06:21.250 It's all about minimizing that error rate. But 00:06:21.250 --> 00:06:23.250 looking at the evolution of these interfaces, 00:06:23.790 --> 00:06:25.689 it brings up an interesting point of friction. 00:06:26.350 --> 00:06:28.730 There's a famous quote from Douglas Engelbart, 00:06:28.889 --> 00:06:31.089 one of the great pioneers of modern computing. 00:06:31.329 --> 00:06:32.750 Oh wait, another one you're talking about. Yeah, 00:06:32.750 --> 00:06:35.870 he said, if ease of use were the only valid criterion, 00:06:36.009 --> 00:06:38.550 people would stick to tricycles and never try 00:06:38.550 --> 00:06:41.389 bicycles. It's a classic. I want to push back 00:06:41.389 --> 00:06:43.709 on this idea a little bit though, because isn't 00:06:43.709 --> 00:06:46.250 the ultimate goal of all this advanced tracking 00:06:46.250 --> 00:06:49.329 and interface design to make technology completely 00:06:49.329 --> 00:06:52.009 frictionless? Shouldn't a good interface feel 00:06:52.009 --> 00:06:55.740 as easy as riding a tricycle? It is a brilliant 00:06:55.740 --> 00:06:58.300 paradox, really, and it gets to the heart of 00:06:58.300 --> 00:07:01.360 advanced HCI. How so? What Engelbert is highlighting 00:07:01.360 --> 00:07:03.579 is the fundamental tension between immediate 00:07:03.579 --> 00:07:07.379 accessibility and long -term capability. A tricycle 00:07:07.379 --> 00:07:09.660 is incredibly easy to use right out of the box, 00:07:09.819 --> 00:07:12.649 right? Sure. Almost zero learning curve. A toddler 00:07:12.649 --> 00:07:16.009 can master it in minutes. Exactly. But its utility 00:07:16.009 --> 00:07:18.670 is permanently capped. You can't navigate rough 00:07:18.670 --> 00:07:21.329 terrain. You can't travel long distances efficiently. 00:07:22.009 --> 00:07:24.529 A bicycle, on the other hand, requires the user 00:07:24.529 --> 00:07:27.449 to learn balance. Which is hard at first. It 00:07:27.449 --> 00:07:30.629 is. It demands a period of friction, of failing 00:07:30.629 --> 00:07:33.649 and falling off. But once the user masters that 00:07:33.649 --> 00:07:37.370 interface, they unlock a vastly more powerful 00:07:37.370 --> 00:07:41.170 tool. Ah, okay, so... If designers only ever 00:07:41.170 --> 00:07:43.910 build tricycle interfaces that require zero effort 00:07:43.910 --> 00:07:46.769 to learn, they are artificially limiting the 00:07:46.769 --> 00:07:50.329 power of the user. Precisely. Sometimes, mastering 00:07:50.329 --> 00:07:53.370 a complex workflow requires a bicycle interface. 00:07:54.069 --> 00:07:55.709 You don't want a commercial flight management 00:07:55.709 --> 00:07:57.569 system controlled by an interface designed to 00:07:57.569 --> 00:07:59.750 be as simple as a smartphone app. Yeah, that 00:07:59.750 --> 00:08:01.879 actually sounds terrifying. Right. You need a 00:08:01.879 --> 00:08:04.120 system whose complexity matches the demands of 00:08:04.120 --> 00:08:06.680 the task, even if that means the human has to 00:08:06.680 --> 00:08:09.319 undergo rigorous training to learn the language 00:08:09.319 --> 00:08:11.680 of that specific machine. OK, that clarifies 00:08:11.680 --> 00:08:13.339 things immensely. You need an interface that 00:08:13.339 --> 00:08:15.300 respects the complexity of the job. Exactly. 00:08:15.439 --> 00:08:17.439 So how do engineers actually figure out what 00:08:17.439 --> 00:08:21.019 humans need for a specific task? How do you systematically 00:08:21.019 --> 00:08:23.480 engineer a human experience without just, you 00:08:23.480 --> 00:08:26.089 know, It's a huge challenge, but it basically 00:08:26.089 --> 00:08:29.509 all hinges on iterative design. Iterative design 00:08:29.509 --> 00:08:32.809 is the engine of HCI. It is a continuous loop. 00:08:33.330 --> 00:08:36.090 You design the user interface, you test it, you 00:08:36.090 --> 00:08:38.570 analyze the empirical results, and then you repeat 00:08:38.570 --> 00:08:42.250 the process. But, and this is key, the entire 00:08:42.250 --> 00:08:45.529 methodology falls apart if you violate one crucial 00:08:45.529 --> 00:08:48.389 rule during the testing phase. You must establish 00:08:48.389 --> 00:08:52.080 exactly who the appropriate user is. Right, because 00:08:52.080 --> 00:08:54.240 testing a complex interface on someone who has 00:08:54.240 --> 00:08:56.320 never used it and will never use it in their 00:08:56.320 --> 00:08:59.039 actual life generates completely invalid data. 00:08:59.440 --> 00:09:02.460 Exactly. You can't test a radar UI on a barista 00:09:02.460 --> 00:09:04.799 and expect to learn anything useful about air 00:09:04.799 --> 00:09:07.370 traffic control. So what does this all mean for 00:09:07.370 --> 00:09:09.450 the researchers? How do they frame this testing? 00:09:09.690 --> 00:09:11.450 Because looking at the source material, there 00:09:11.450 --> 00:09:14.049 are three major design philosophies that dictate 00:09:14.049 --> 00:09:16.649 how these interfaces are built. Activity theory, 00:09:17.129 --> 00:09:19.889 user -centered design, and value -sensitive design. 00:09:20.070 --> 00:09:21.970 Yeah, let's break those down, starting with activity 00:09:21.970 --> 00:09:24.889 theory. This philosophy gives designers a framework 00:09:24.889 --> 00:09:27.549 for defining the context in which the human interacts 00:09:27.549 --> 00:09:30.549 with the computer. It focuses heavily on the 00:09:30.549 --> 00:09:33.110 action itself. Like the actual task they are 00:09:33.110 --> 00:09:36.679 doing. Right. What is the human trying to achieve? 00:09:37.179 --> 00:09:39.580 And what environmental factors influence that 00:09:39.580 --> 00:09:43.519 action? It is very pragmatic. And then user -centered 00:09:43.519 --> 00:09:46.740 design, or UCD, shifts the focus slightly. It 00:09:46.740 --> 00:09:49.299 does. It places the user, rather than the task, 00:09:49.480 --> 00:09:52.100 at the absolute center of the process. And UCD 00:09:52.100 --> 00:09:54.519 relies heavily on ethnographic studies, which 00:09:54.519 --> 00:09:57.100 I find fascinating. The designers act almost 00:09:57.100 --> 00:09:59.259 like anthropologists. That's a great way to put 00:09:59.259 --> 00:10:01.720 it. They don't just bring people into a sterile 00:10:01.720 --> 00:10:03.639 laboratory with a two -way mirror. They go out 00:10:03.639 --> 00:10:06.500 into the field. They observe the actual environment 00:10:06.500 --> 00:10:08.740 where the user operates, whether that's a noisy 00:10:08.740 --> 00:10:11.440 emergency room or a chaotic trading floor, to 00:10:11.440 --> 00:10:13.820 see how humans behave naturally when interacting 00:10:13.820 --> 00:10:16.100 with the system. Because they recognize that 00:10:16.100 --> 00:10:19.659 humans are complex social creatures. An interface 00:10:19.659 --> 00:10:22.019 that works perfectly in a quiet, brightly lit 00:10:22.019 --> 00:10:25.759 lab might be completely unusable in a dimly lit, 00:10:25.919 --> 00:10:27.879 high stress environment. Yeah, that makes perfect 00:10:27.879 --> 00:10:30.759 sense. understanding leads directly into the 00:10:30.759 --> 00:10:33.639 third philosophy value sensitive design, or VSD. 00:10:34.759 --> 00:10:37.059 This methodology expands the scope significantly. 00:10:37.159 --> 00:10:39.700 How much bigger are we talking? Well, it dictates 00:10:39.700 --> 00:10:42.460 that an interface must account not just for the 00:10:42.460 --> 00:10:44.519 individuals who use the technology directly, 00:10:45.139 --> 00:10:48.639 but for anyone who the design influences indirectly. 00:10:48.840 --> 00:10:51.779 Wow, that is a massive distinction. You are essentially 00:10:51.779 --> 00:10:54.200 designing for the ripple effect. Exactly. It 00:10:54.200 --> 00:10:56.940 requires a holistic approach utilizing three 00:10:56.940 --> 00:10:59.059 distinct kinds of investigations. OK, let's hear 00:10:59.059 --> 00:11:01.600 them. First, conceptual investigations, where 00:11:01.600 --> 00:11:04.360 researchers try to articulate the different human 00:11:04.360 --> 00:11:07.779 values at stake and identify potential conflicts. 00:11:08.139 --> 00:11:10.200 Can you give me an example of that? Sure. Like 00:11:10.200 --> 00:11:12.940 balancing a user's desire for frictionless sharing 00:11:12.940 --> 00:11:16.019 on a social app with a non -user's right to privacy. 00:11:16.139 --> 00:11:19.639 Ah, okay, that's a tough balance. It is. Second, 00:11:19.840 --> 00:11:21.840 you have empirical investigations which involve 00:11:21.840 --> 00:11:24.259 qualitative or quantitative studies to see how 00:11:24.259 --> 00:11:26.500 these values actually play out in real -world 00:11:26.500 --> 00:11:29.779 use. So testing the theories. Right. And third, 00:11:30.200 --> 00:11:32.500 technical investigations, which evaluate how 00:11:32.500 --> 00:11:34.799 the underlying architecture of the system either 00:11:34.799 --> 00:11:37.500 supports or undermines those identified values. 00:11:38.379 --> 00:11:40.679 VSD attempts to ensure the technology serves 00:11:40.679 --> 00:11:43.259 the broader human ecosystem. But here is the 00:11:43.259 --> 00:11:46.399 inherent problem with trying to engineer a flawless 00:11:46.399 --> 00:11:50.139 system around human beings. We are profoundly 00:11:50.139 --> 00:11:53.139 messy. Very true. Methodologies like VSD are 00:11:53.139 --> 00:11:56.620 great in theory, but humans are not purely logical 00:11:56.620 --> 00:11:59.500 machines. We are irrational, we are emotional, 00:11:59.500 --> 00:12:02.600 and we carry deeply ingrained social biases. 00:12:03.419 --> 00:12:06.019 How does an interface handle that? This is where 00:12:06.019 --> 00:12:08.580 things get really wild. This is where the fields 00:12:08.580 --> 00:12:10.820 of social computing and effective computing come 00:12:10.820 --> 00:12:13.220 into play. Okay, social computing first. Right. 00:12:13.600 --> 00:12:15.960 Social computing examines the interactive and 00:12:15.960 --> 00:12:18.100 collaborative behavior between technology and 00:12:18.100 --> 00:12:21.179 people, drawing heavily from sociology and psychology. 00:12:21.580 --> 00:12:24.179 And it exposes some incredible quirks in human 00:12:24.179 --> 00:12:26.919 nature. Like what? Well, in one prominent study, 00:12:27.399 --> 00:12:29.379 users were asked to evaluate different computers. 00:12:29.629 --> 00:12:31.590 The researchers found that people consistently 00:12:31.590 --> 00:12:34.110 expected a computer given a man's name to cost 00:12:34.110 --> 00:12:36.350 more than a machine given a woman's name. Wait, 00:12:36.370 --> 00:12:39.269 really? We are taking a literal box of plastic, 00:12:39.490 --> 00:12:42.490 wire, and silicon, and simply because of a label, 00:12:42.570 --> 00:12:45.629 we project societal gender biases onto a motherboard. 00:12:45.809 --> 00:12:48.529 Yes, exactly. It really highlights how we anthropomorphize 00:12:48.529 --> 00:12:51.730 our tools. And supposedly, we also generally 00:12:51.730 --> 00:12:53.950 perceive our interactions with computers more 00:12:53.950 --> 00:12:57.059 negatively than with humans. even if the computer's 00:12:57.059 --> 00:12:59.899 output and behavior are identical to what a human 00:12:59.899 --> 00:13:02.580 would do. Yeah, we just inherently harbor a subtle 00:13:02.580 --> 00:13:05.740 hostility toward the machine. We approach the 00:13:05.740 --> 00:13:08.580 dialogue with an inherent bias. That's so strange. 00:13:08.779 --> 00:13:12.179 It is, but that is exactly why effective computing, 00:13:12.480 --> 00:13:15.519 which is the study of emotions within HCI, is 00:13:15.519 --> 00:13:18.269 expanding so rapidly. Researchers are developing 00:13:18.269 --> 00:13:21.269 systems that can detect, process, and react to 00:13:21.269 --> 00:13:23.730 human emotional states. How do they even do that? 00:13:23.730 --> 00:13:25.450 They do this through effect detection channels. 00:13:25.570 --> 00:13:27.450 Here's where it gets really interesting, because 00:13:27.450 --> 00:13:30.389 that means using physiological data. Yes. Things 00:13:30.389 --> 00:13:34.210 like monitoring an ECG to analyze a user's financial 00:13:34.210 --> 00:13:37.350 decision -making process under stress, or using 00:13:37.350 --> 00:13:39.870 eye tracking and pupil dilation to understand 00:13:39.870 --> 00:13:41.990 how organizational knowledge is being absorbed 00:13:41.990 --> 00:13:44.110 in a corporate setting. It's incredible tech. 00:13:44.230 --> 00:13:46.759 We're projecting our human biases onto machines, 00:13:46.820 --> 00:13:49.679 but the machines are quietly reading our biological 00:13:49.679 --> 00:13:52.539 stress signals and micro expressions. But from 00:13:52.539 --> 00:13:55.179 a technical standpoint, the massive hurdle here 00:13:55.179 --> 00:13:57.899 is how the computer interprets that biological 00:13:57.899 --> 00:14:00.080 data. Right, because an elevated heart rate could 00:14:00.080 --> 00:14:02.620 mean the user is frustrated by a software bug 00:14:02.620 --> 00:14:04.639 or... Could mean they just drank a large coffee. 00:14:04.759 --> 00:14:07.019 Exactly. So how do they solve that? To solve 00:14:07.019 --> 00:14:09.899 this, HCI relies on the concept of ontology. 00:14:10.120 --> 00:14:13.000 Okay, how does ontology actually work in this 00:14:13.000 --> 00:14:16.289 context? So in information science... There is 00:14:16.289 --> 00:14:19.049 a semantic gap between human understanding and 00:14:19.049 --> 00:14:21.090 machine understanding. Meaning they don't speak 00:14:21.090 --> 00:14:23.590 the same language. Exactly. Humans understand 00:14:23.590 --> 00:14:27.450 context, nuance, and emotion. Machines understand 00:14:27.450 --> 00:14:31.590 binary data. Ontology is the creation of a formal, 00:14:31.909 --> 00:14:33.950 highly structured dictionary of domain -specific 00:14:33.950 --> 00:14:36.429 knowledge. Okay, a dictionary. Right. It maps 00:14:36.429 --> 00:14:39.590 specific physiological inputs, like a furrowed 00:14:39.590 --> 00:14:41.990 brow combined with an erratic mouse movement, 00:14:42.409 --> 00:14:45.889 to a defined concept like user frustration. I 00:14:45.889 --> 00:14:47.850 see. It provides the architectural bridge that 00:14:47.850 --> 00:14:49.929 translates the messy reality of human panic, 00:14:50.169 --> 00:14:53.090 joy, or confusion into actionable data the computer 00:14:53.090 --> 00:14:55.870 can use to dynamically alter its interface. So 00:14:55.870 --> 00:14:58.490 we've moved from typing commands on a clunky 00:14:58.490 --> 00:15:02.269 keyboard to having a machine use ontology to 00:15:02.269 --> 00:15:04.789 translate our heart rates into interface adjustments. 00:15:04.990 --> 00:15:06.950 It's a huge leap. And the logical progression 00:15:06.950 --> 00:15:10.129 here is the literal merging of the human and 00:15:10.129 --> 00:15:12.649 machine environments. The boundary is completely 00:15:12.649 --> 00:15:15.250 blurring across the XR continuum. Oh definitely. 00:15:15.470 --> 00:15:18.590 The shift from 2D screens to spatial computing 00:15:18.590 --> 00:15:20.889 fundamentally alters the interaction loop we 00:15:20.889 --> 00:15:22.549 talked about earlier. Because we aren't just 00:15:22.549 --> 00:15:24.750 looking at a screen anymore. Right. Augmented 00:15:24.750 --> 00:15:27.809 reality, or AR, integrates digital information 00:15:27.809 --> 00:15:29.789 into your physical environment, enhancing what 00:15:29.789 --> 00:15:33.090 you see. Virtual reality completely replaces 00:15:33.090 --> 00:15:35.429 the physical environment, placing the user in 00:15:35.429 --> 00:15:37.889 a fully immersive digital space. Which has to 00:15:37.889 --> 00:15:39.970 drastically change how we study user presence 00:15:39.970 --> 00:15:42.110 and cognitive load. Absolutely. And then mixed 00:15:42.110 --> 00:15:44.620 reality blends both. allowing for real -time 00:15:44.620 --> 00:15:46.519 interaction with physical and digital objects 00:15:46.519 --> 00:15:50.720 simultaneously. Extended reality, or XR, encompasses 00:15:50.720 --> 00:15:53.139 this entire spectrum. We are moving from gazing 00:15:53.139 --> 00:15:56.679 at a flat screen to utilizing 3D spatial awareness 00:15:56.679 --> 00:15:58.919 and kinetic feedback. But as these immersive 00:15:58.919 --> 00:16:00.500 environments evolve, the systems running them 00:16:00.500 --> 00:16:03.299 are becoming autonomous. Yes. The AI element 00:16:03.299 --> 00:16:06.179 is huge. When we look at human -AI interaction, 00:16:06.399 --> 00:16:08.980 the dialogue changes again. You aren't just commanding 00:16:08.980 --> 00:16:11.399 a tool, you are collaborating with an agent. 00:16:11.500 --> 00:16:13.419 Which is why a major focus of current research 00:16:13.419 --> 00:16:17.019 is XAI, or Explainable AI. Explainable AI. Yeah. 00:16:17.279 --> 00:16:19.480 If a human is going to collaborate with an AI, 00:16:19.779 --> 00:16:21.879 especially in a high -stakes environment like 00:16:21.879 --> 00:16:25.200 medical diagnostics or autonomous driving, the 00:16:25.200 --> 00:16:27.899 AI's outputs cannot come from a black box. Right. 00:16:27.899 --> 00:16:30.240 I need to know why it made that decision. Exactly. 00:16:30.659 --> 00:16:33.740 The interaction loop requires transparency. XAI 00:16:33.740 --> 00:16:36.179 ensures that the system's decision -making process 00:16:36.179 --> 00:16:38.980 is interpretable to the human user, which is 00:16:38.980 --> 00:16:41.080 the only way to establish and maintain trust. 00:16:41.379 --> 00:16:43.980 But the ultimate interface, the absolute final 00:16:43.980 --> 00:16:47.179 frontier of HCI, strips away the screens, the 00:16:47.179 --> 00:16:49.860 spatial holograms, and the voice commands entirely. 00:16:49.940 --> 00:16:52.360 Oh, you're talking about BCIs. Yes, brain -computer 00:16:52.360 --> 00:16:55.179 interfaces. We are talking about a direct communication 00:16:55.179 --> 00:16:57.539 pathway between a wired brain and an external 00:16:57.539 --> 00:16:59.899 device. And there is a vital distinction here, 00:16:59.980 --> 00:17:02.860 according to the text. A BCI is fundamentally 00:17:02.860 --> 00:17:05.180 different from a medical neuromodulation device, 00:17:05.400 --> 00:17:07.759 like a pacemaker for the brain. Right, because 00:17:07.759 --> 00:17:10.700 a BCI allows for bidirectional information flow. 00:17:11.359 --> 00:17:13.119 Bidirectional? Yeah, that bidirectional nature 00:17:13.119 --> 00:17:15.599 is what makes it a true interface. Information 00:17:15.599 --> 00:17:17.740 is not just being read from the neural pathways 00:17:17.740 --> 00:17:20.920 to control a cursor or a robotic limb. Data is 00:17:20.920 --> 00:17:23.160 being sent back into the brain. That is mind 00:17:23.160 --> 00:17:25.880 -blowing. Literally. It is used to research, 00:17:26.359 --> 00:17:30.059 map, augment, or repair human cognitive and sensory 00:17:30.059 --> 00:17:32.799 motor functions. The loop of interaction we discussed 00:17:32.799 --> 00:17:34.839 earlier is no longer happening through external 00:17:34.839 --> 00:17:37.980 physical channels. It is happening directly across 00:17:37.980 --> 00:17:40.440 biological synapses. Think about that progression 00:17:40.440 --> 00:17:42.839 for a moment for you listening. We went from 00:17:42.839 --> 00:17:44.680 talking to the computer through a window like 00:17:44.680 --> 00:17:47.569 a desktop monitor. Then, with XR, we stepped 00:17:47.569 --> 00:17:50.470 inside the room with the computer. And now, with 00:17:50.470 --> 00:17:53.569 BCIs, we are wiring the computer directly into 00:17:53.569 --> 00:17:55.789 our nervous system. It's moving so fast. But 00:17:55.789 --> 00:17:57.930 with technology integrating this intimately with 00:17:57.930 --> 00:18:00.769 our biology and our daily lives, two massive 00:18:00.769 --> 00:18:03.549 vulnerabilities become glaringly obvious. First, 00:18:03.569 --> 00:18:06.589 who gets left behind? And second, how do we secure 00:18:06.589 --> 00:18:08.329 the connection? Both of those vulnerabilities 00:18:08.329 --> 00:18:11.430 are critical areas of study right now. Regarding 00:18:11.430 --> 00:18:14.250 who gets left behind, the field focuses heavily 00:18:14.250 --> 00:18:17.109 on accessibility. Okay, so designing for disabilities. 00:18:17.430 --> 00:18:20.049 Right. The goal is to design digital experiences 00:18:20.049 --> 00:18:23.089 that are inclusive for individuals with visual, 00:18:23.470 --> 00:18:26.589 motor, or cognitive disabilities. But the research 00:18:26.589 --> 00:18:30.069 yields a fascinating universal truth. Which is? 00:18:30.210 --> 00:18:32.509 Accessible design doesn't just accommodate specific 00:18:32.509 --> 00:18:35.349 groups. It inherently enhances usability for 00:18:35.349 --> 00:18:38.410 all users. It is the digital equivalent of the 00:18:38.410 --> 00:18:40.450 sidewalk curb cut. Oh, that's a perfect example. 00:18:40.630 --> 00:18:42.589 Right. The slope in the concrete was originally 00:18:42.589 --> 00:18:45.109 mandated and designed for wheelchairs, but it 00:18:45.109 --> 00:18:47.630 ends up helping anyone pushing a stroller, dragging 00:18:47.630 --> 00:18:50.849 a suitcase, or riding a bicycle. Universal design 00:18:50.849 --> 00:18:53.309 just creates a cleaner, more logical architecture 00:18:53.309 --> 00:18:55.569 for everyone. Exactly. But then we pivot to the 00:18:55.569 --> 00:18:58.130 other vulnerability, which is security. This 00:18:58.130 --> 00:19:00.950 is the domain of HCI -SEC or security interactions 00:19:00.950 --> 00:19:03.670 and the timeline here is surprising. It really 00:19:03.670 --> 00:19:07.369 is. HCI as a field has deep roots going all the 00:19:07.369 --> 00:19:10.630 way back to the 1970s at places like Xerox PARC. 00:19:11.309 --> 00:19:13.970 But security usability is a completely nascent 00:19:13.970 --> 00:19:17.559 field. It is surprisingly new. Interest in HC 00:19:17.559 --> 00:19:19.960 -ISAC has really only tracked what the recent 00:19:19.960 --> 00:19:23.059 explosion of internet security as a broad mainstream 00:19:23.059 --> 00:19:26.339 concern. Interesting. The aim is to improve the 00:19:26.339 --> 00:19:28.980 usability of security features for the end user, 00:19:29.619 --> 00:19:32.059 because historically, as we all know, security 00:19:32.059 --> 00:19:35.769 usability is... notoriously terrible. We all 00:19:35.769 --> 00:19:38.630 experience this constantly. Every corporate portal 00:19:38.630 --> 00:19:41.529 or banking app seems to require an overly complex 00:19:41.529 --> 00:19:44.269 ritual of authenticators, special characters, 00:19:44.490 --> 00:19:47.390 and expiring passwords just to log in, and it 00:19:47.390 --> 00:19:49.849 constantly breaks the workflow. It's exhausting. 00:19:50.190 --> 00:19:52.750 Why are security interfaces always the clunkiest 00:19:52.750 --> 00:19:55.990 part of any software ecosystem? There are several 00:19:55.990 --> 00:19:58.220 structural reasons for that. Security features 00:19:58.220 --> 00:20:00.420 are frequently added as a casual afterthought, 00:20:00.680 --> 00:20:03.180 or they're hastily patched in to address a newly 00:20:03.180 --> 00:20:05.579 discovered vulnerability, rather than being built 00:20:05.579 --> 00:20:07.160 into the foundational design of the interface. 00:20:07.220 --> 00:20:10.119 But they're band -aids. Basically. They address 00:20:10.119 --> 00:20:12.779 incredibly complex cryptographic use cases, but 00:20:12.779 --> 00:20:14.720 rarely provide a software wizard to help the 00:20:14.720 --> 00:20:16.960 user navigate that complexity. And the designers. 00:20:17.359 --> 00:20:20.240 Well, that's another issue. The interface designers 00:20:20.240 --> 00:20:23.099 themselves often lack an understanding of the 00:20:23.099 --> 00:20:25.930 underlying security concepts. Or, conversely, 00:20:26.230 --> 00:20:28.569 the people designing the security interface aren't 00:20:28.569 --> 00:20:30.910 usability experts at all. They are just the back 00:20:30.910 --> 00:20:32.849 -end application developers who wrote the toad. 00:20:33.029 --> 00:20:36.470 Which creates a massive point of failure. Because 00:20:36.470 --> 00:20:38.670 what good is a state -of -the -art cryptographic 00:20:38.670 --> 00:20:41.549 security system if the user is so frustrated 00:20:41.549 --> 00:20:43.890 by the clunky interface that they just write 00:20:43.890 --> 00:20:46.190 their master password on a sticky note and slap 00:20:46.190 --> 00:20:49.250 it on the side of their monitor? Exactly. the 00:20:49.250 --> 00:20:51.250 human element will always seek the path of least 00:20:51.250 --> 00:20:53.809 resistance. An unusable security feature isn't 00:20:53.809 --> 00:20:56.210 just annoying, it is essentially no security 00:20:56.210 --> 00:20:59.230 at all. It compromises the entire architecture. 00:20:59.549 --> 00:21:01.390 This raises an important question, right? Right. 00:21:01.549 --> 00:21:03.670 Because it completely undermines the system. 00:21:04.009 --> 00:21:06.690 The strongest lock in the world is useless if 00:21:06.690 --> 00:21:09.650 the process of using the key is so painful that 00:21:09.650 --> 00:21:13.190 the user just props the door open. Wow. For you 00:21:13.190 --> 00:21:15.589 listening, think about the journey we just took. 00:21:16.009 --> 00:21:18.750 We started with the physical dials and blaring 00:21:18.750 --> 00:21:21.809 alarms of a nuclear power plant from the 1970s. 00:21:21.849 --> 00:21:24.970 We explored the paradox of tricycles versus bicycles 00:21:24.970 --> 00:21:28.029 and how engineers try to map human values into 00:21:28.029 --> 00:21:30.750 code. We covered a lot of ground. We really did. 00:21:31.029 --> 00:21:33.190 We moved through the psychological quirks of 00:21:33.190 --> 00:21:35.450 social computing, where we project human biases 00:21:35.450 --> 00:21:38.470 onto inanimate boxes, while the machines use 00:21:38.470 --> 00:21:41.109 ontology to translate our physiological stress 00:21:41.109 --> 00:21:43.349 into data. And then we ended up looking at a 00:21:43.349 --> 00:21:46.859 future where bi -directional interfaces are literally 00:21:46.859 --> 00:21:49.460 wiring our nervous systems into the digital world. 00:21:49.900 --> 00:21:51.799 Human -computer interaction isn't just about 00:21:51.799 --> 00:21:54.079 making an app look pretty. It is the invisible 00:21:54.079 --> 00:21:56.839 architecture of our modern lives. It dictates 00:21:56.839 --> 00:21:59.460 how we work, how we communicate, and how we exist 00:21:59.460 --> 00:22:01.839 in the 21st century. It really is the defining 00:22:01.839 --> 00:22:04.460 language of our era. And as we wrap up this deep 00:22:04.460 --> 00:22:05.960 dive, I want to leave you with a final thought 00:22:05.960 --> 00:22:08.539 to mull over, building on the reality of brain 00:22:08.539 --> 00:22:10.680 -computer interfaces and effective computing. 00:22:10.920 --> 00:22:13.339 Okay, let's hear it. We know that BCIs allow 00:22:13.339 --> 00:22:16.680 for true bidirectional flow, and we know that 00:22:16.680 --> 00:22:19.119 systems are rapidly learning to detect and process 00:22:19.119 --> 00:22:22.039 our emotions. So what happens when a machine 00:22:22.039 --> 00:22:24.579 cannot just read your emotional state to understand 00:22:24.579 --> 00:22:27.720 you, but potentially write or alter a feeling 00:22:27.720 --> 00:22:30.440 directly into your neural pathways? When the 00:22:30.440 --> 00:22:32.380 open -ended dialogue between human and computer 00:22:32.380 --> 00:22:34.759 happens entirely inside your own mind without 00:22:34.759 --> 00:22:37.400 a screen or a spoken word, where does the machine 00:22:37.400 --> 00:22:40.240 end and where do you begin? That is a staggering 00:22:40.240 --> 00:22:43.200 thought to leave on. A literal dialogue inside 00:22:43.200 --> 00:22:45.140 the mind. Thank you so much for joining us on 00:22:45.140 --> 00:22:47.359 this deep dive into human -computer interaction. 00:22:47.599 --> 00:22:49.480 We hope you look at the interfaces around you 00:22:49.480 --> 00:22:51.160 and the silent conversations you're having with 00:22:51.160 --> 00:22:53.700 them every single day, a little differently from 00:22:53.700 --> 00:22:54.980 now on. Until next time.