Name: ハーバードのコンピュータ・サイエンスの教授だけど質問ある？ | Tech Support | WIRED Japan
Uploaded: 2025-03-27T08:45:45.000Z
Duration: 14 min 27 s
Description: VoiceTubeの動画で発音を聞きながら英語表現を覚えよう！学べる英語：

I teach computer science at Harvard, and I'm here today to answer your questions from Twitter.

Well, the short answer really is distributed computing, which is to say that Google and Bing and other such search engines, they don't just have one server, and they don't even have just one really big server.

Rather, they have hundreds, thousands, probably hundreds of thousands or more servers nowadays around the world.

And so when you and I go to Google or Bing and maybe type in a word to search for, like cats, it's quite possible that when you hit Enter, and that keyword, like cats, is sent over the internet to Google or to Bing, it's actually spread out ultimately across multiple servers, some of which are grabbing the first 10 results, some of which are grabbing the next 10 results, the next 10 results, so that you see just one collection of results, but a lot of those ideas, a lot of those search results came from different places.

And this eliminates what could potentially be a bottleneck of sorts if all of the information you needed had to come from one specific server that might very well be busy when you have that question.

Nick asks, will computer programming jobs be taken over by AI within the next five to 10 years?

This is such a frequently asked question nowadays, and I don't think the answer will be yes.

And I think we've seen evidence of this already in that early on when people were creating websites, they were literally writing out code in a language called HTML by hand.

But then, of course, software came along, tools like Dreamweaver that you could download on your own computer that would generate some of that same code for you.

More recently, though, now you can just sign up for websites like Squarespace and Wix and others whereby click, click, click, and the website is generated for you.

So I dare say, certainly in some domains, that AI is really just an evolution of that trend, and it hasn't put humans out of business as much as it has made you and I much more productive.

AI, I think, and the ability soon to be able to program with natural language is just gonna enhance what you and I can already do logically, but much more mechanically.

And I think, too, it's worth considering that there's just so many bugs or mistakes in software in the world, and there's so many features that humans wish existed in products present and future that our to-do list, so to speak, is way longer than we'll ever have time to finish in our lifetimes.

And so I think the prospect of having an artificial intelligence boost our productivity and work alongside us, so to speak, as we try to solve problems, is just gonna mean that you and I and the world together can solve so many more problems and move forward together at an even faster rate.

All right, next up, Sophia, who asks, how do microchips even work?

Well, here, for instance, we have a whole bunch of microchips on what's called a logic board, or sometimes known as a motherboard.

There's a lot of ports that you might be familiar with, for instance, like here's some ports for audio, here's some ports for networking, here's some ports for USB and some other devices as well.

And those ports, meanwhile, are connected to lots of different chips on this board that know how to interpret the signals from those ports.

And perhaps the biggest chip on this motherboard tends to be this thing here called the CPU, or the central processing unit, which is really the brains of the computer.

And what you can't necessarily quite see, because most of this is actually paint and not traces, but if I flip this around, you'll actually see in the right light and with the right angle a whole bunch of traces running up, down, left, and right on this logic board that's connecting all of these various microchips.

And by trace, I mean a tiny little wire that's been etched into the top or the bottom of this circuit board that connects two parts thereon.

Now, what might these microchips be doing?

Well, again, they might be simply interpreting signals that are coming in from these ports.

Two, they might be performing mathematical operations, doing something with those signals in order to convert input into output, or they might just be storing information ultimately.

In fact, there's all different types of memory on a logic board like this, be it RAM or ROM or the like, and so some of those chips might very well be storing information.

For as long as the computer's plugged in, or in some cases, depending on the device, even when the power goes off.

So if anyone can learn coding, what do computer scientists do for four years in university?

Typically, in an undergraduate program in computer science or computer engineering or a similar field, someone spends much more time learning about the field itself than about programming specifically.

So as such, you might study not only a bit of programming, but also mathematics, certain fundamentals that transcend the particular classes you might have taken in middle school or high school, but that can be used to solve grander real-world problems.

You might learn something about networks, how you can send information from point A to point B.

You might learn about graphics, how you can display things on the screen or even create interactive animations or the like.

You might learn how to leverage certain ideas from mathematics and other fields to implement your very own artificial intelligence nowadays, whereby you use probability and statistics and information more generally to try to predict what a intelligent individual, or in this case, computer, might say in response to a question.

So computer science itself is a very broad field and programming is really just a tool that you tend to learn along the way.

For Maya Shelby, how do zeros and ones turn into the internet?

Well, I think the simplest answer there is that the internet is built upon layers and layers and layers of ideas.

And if we start at the lowest of those levels, zeros and ones, you have something called binary, where zeros and ones can be used to represent any other numbers as well.

And if we use more and more zeros and ones, more and more binary digits or bits, so to speak, we can count up higher and higher and higher.

And then if you and I agree that, all right, well, let's not just use these patterns of zeros and ones to represent numbers.

What if we reserve some of these patterns to represent letters of like the English alphabet?

And then maybe you and I can decide to reserve certain patterns of zeros and ones to represent colors like red and green and blue and combinations thereof.

Well, once we have the ability to represent colors, we could then represent whole pictures.

Because what's a picture on your phone or a computer screen?

Well, it's really just a grid of dots, each of which has its own color.

So this is all to say that even if we start at this lowest level of just zeros and ones, so long as you and I and all of the devices we use agree to follow some standard like this, we can build these layers and layers of abstraction, so to speak, on top of one another, until finally you and I come up with a pattern of zeros and ones that represents send this piece of information from me over there.

And thus, we have something like the internet.

Mojinbu asks, can someone that knows computer science explain to me why computers use binary coding and not trinary, when trinary is supposed to be faster?

So it's not necessarily the case that a trinary system, which would use three symbols, for instance, zero, one, and two, would necessarily be faster than binary because binary, using just zero and one, tends to be simpler to implement and also more robust to potential errors.

Or if you're familiar with voltage levels, like in a battery, it's very easy for a computer to distinguish something for like zero volts or three volts, but it gets a little harder if we try to draw the line somewhere in between because there's just a higher probability that a computer might mistake a voltage level like 1.5 in the middle as maybe being a little closer to off than on or to on than off.

Here, too, is where even though there might be mathematical efficiencies and real world efficiencies to using trinary, otherwise known as ternary, like a zero, a one, and a two digit, instead of just zeros and ones, it turns out because our world runs on electricity nowadays and there's so much momentum behind binary that it just tends to be a net positive.

Rachel P95 asks, why is every Windows solution, have you tried restarting, and why does that always work?

So that's a very heavy-handed solution to what are typically just bugs or mistakes in software, for instance, Windows in this case.

Restarting a computer just starts everything from scratch.

So all of the computer's short-term memory is lost and everything starts in pristine condition, which is to say that it starts in exactly the way that the programmers at Microsoft intended without potentially the distractions of the computer being in some weird state or condition that the programmers just didn't anticipate.

Maybe you clicked on some buttons in a weird order, maybe you opened a strange file, but maybe you got the computer into a state that just wasn't programmed for properly.

Jason Whitmer now asks, what's the best operating system?

Well, this is one of these questions in computing we would call a religious question since it evokes a religious debate as to which might be best.

Of course, among the most popular operating systems out there are Windows and Mac OS, but there's also one you might not have heard of, which is called Linux, which is actually very omnipresent in the enterprise world.

So many of today's servers actually run Linux.

So many of today's desktops or laptops, though, run Windows or Mac OS.

Now, that's not to say you couldn't run all of those operating systems in different contexts, and some of us do actually run Linux on our own computers, so a lot of it really boils down to personal preference.

I wouldn't even say that there's one best operating system, but there tend to be correlations between the operating systems people use and the applications they have in mind.

So Windows, for instance, is so popular in the world of PCs and desktops and laptops.

Mac OS is, to some extent, particularly in academia and certain countries, but not necessarily on the same scale.

Linux, by contrast, is, again, very much used heavily in the server-side industry, but so is Windows as well.

So a lot of the choice for operating systems sometimes comes from just what's most appropriate, what's most popular, what's most supportive, but some of it comes, too, from just personal preference of the engineer, maybe the first engineer that you hire to make one of those decisions.

So it's more about what's best for you and not so much best in general.

Next, Julio Magnifico asks, why aren't computers getting cheaper?

Well, computers, or at least computer parts inside of computers, do tend to get cheaper.

The catch is that your expectations and my expectations just keep rising.

We want our phones, our laptops, our desktops to do more and more in the way of the software that they run, the games that we use, and just how quickly they perform for us.

So even though some of those parts are getting less expensive, you and I want them to do more and more and be faster and larger in quantity.

And so as a result, I dare say that the price isn't going down as far as you might hope.

That said, nowadays, you can get, for the same amount of money from yesteryear, much, much more in the way of computing power.

So arguably, it's working to our benefit in some cases.

Next up, from Darrow Nabila, can someone explain cloud computing to me, like a five-year-old?

Cloud computing is essentially you using someone else's servers that someone is paying to rent, for instance, or timeshare.

So this isn't really a new idea or a new technology.

Rather, it's a better branding of a technique that's been used for years, not just in the computer world, but in the real world as well, whereby someone like Google or Microsoft or Apple or others nowadays might be able to afford lots and lots and lots of servers and then make those servers available in part to me, to you, and many other customers as well.

Well, Marcus asks, how does computer memory work?

Think of computer memory as really being driven by a whole bunch of switches that can either be turned on and off.

So for instance, if I take this here light switch, which is currently off, I could simply say that this switch here is representing the number zero in binary.

But if I turn the switch on, well, now I can say that I'm representing the number one.

Now, of course, I can only count as high as zero to one with a single light switch, but what if I bring over a second light switch like this one here?

If we started zero in this way, turn on this switch first and claim that it's one, let me now be more creative and turn this one off and this one on, and now claim this is how a computer's memory could represent the number two.

And now if I turn this switch back on, giving me a fourth pattern, this is how I might represent the number three.

Now, of course, if we add more and more of these switches, more and more of these light bulbs, we can count even higher than three.

And indeed, that's what a computer's memory is ultimately doing.

It's using lots and lots of little tiny switches, otherwise known as transistors, to turn the flow of electricity on and off, and then it's got other types of hardware, called, for instance, capacitors, that have a capacity to hold on to some of that electricity, just like the light bulb there being on.

All right, next, Donnie asks, how do you explain Web 3 to people?

So Web 3, like Web 2, and retrospectively Web 1, are really just buzzwords that describe sort of phases of the internet or the World Wide Web as you and I know it.

字幕リスト動画再生

ハーバードのコンピュータ・サイエンスの教授だけど質問ある？ | Tech Support | WIRED Japan

ultimately

bunch

content

grab