The Perfection of Artificial Intelligence

The text below is a literal transcription of the spoken video text.

This week, I watched two reports about artificial intelligence, and I wanted to say something about that. The first was a report about the future—let’s say it was a positive artificial intelligence figure who envisions that this intelligence will conquer the world. She proclaimed that we must prepare for the fact that, in 2025, a large part of office work, office labor, can be taken over by artificial intelligence. Watch video on YouTube

This week, I watched two reports about artificial intelligence, and I wanted to say something about that. The first was a report about the future—let’s say it was a positive artificial intelligence figure who envisions that this intelligence will conquer the world. She proclaimed that we must prepare for the fact that, in 2025, a large part of office work, office labor, can be taken over by artificial intelligence. I was quite surprised by that because, well, we also work with all sorts of systems to simplify administration, but the most characteristic thing about them is that they are illogical, that they don’t work, that they crash, that they create all sorts of unexpected problems, that those who are supposed to support them don’t know how either, and so on and so on. So that is how we know the "perfection" of artificial intelligence in the administrative field.

Maybe it’s the case that we don’t get the quality that actually exists; I could imagine that. That in reality, the development is much further along, but that it is not available for the—well, let’s say the little people—but for ministries, military departments, and the like, for example. I don’t know. But I had to laugh a little about it because I thought: yes, look, you then have to imagine that not only administration as we know it is handled by a machine, but that it also receives emails, reads them, responds to them, creates quotes, receives responses, decides whether or not to engage with them… That whole job that an entrepreneur does, that would be taken over by artificial intelligence? Well, think about it yourself. I don’t know, I have to laugh a little about it and think: well, this is an overestimation of the possibilities of, let’s say, “have I got something for you.”

But it is presented very seriously. In 2025, it will be the case that a large part of this kind of work can be taken over by the machine. That was one thing. Well, there were more things. The following year, the robot would be available for purchase; it would cost €80,000 or dollars or maybe even 100,000, so not yet available to everyone right away. But in the future, everything becomes cheaper, so you can expect that at some point, you can buy a butler who does who-knows-what for you. Maybe it will even drive to the store with the car—that’s how it is presented—and I cannot take it seriously.

That’s one thing. The other was that I saw a report about someone giving instructions on how to write “the best piece in the world” on a particular subject. You do have to collect quite a few commands for the machine; you have to specify how you want the exposition, what exactly the subject is, with possible subtopics, which words you want to use, how often they should appear, what style it should be—should it be scientific, or should it be more artistic? You can specify all of that, and then you can also say: I want a detailed bibliography, I want people to see what "I"—we’ll talk about that later—have used in terms of literature to arrive at this best result in the world.

So you have to enter all of that, but then the machine starts writing, and according to this report, it produces an exceptionally high-quality article. Well, okay, let’s assume—no example was given—but let’s assume that something is indeed written that has some content and value, maybe even of high quality. Then you still have to ask yourself what this actually means.

Imagine, you have to write a paper for your studies on a particular subject. Of course, the idea is that you use your own knowledge, that you supplement that knowledge with literature research—at least, that’s how I see it. Just imagine how much effort that takes, how much you have to put into it, how many hours it costs. Nowadays, of course, you no longer go to the library, but you go on the internet, and you search, and you find this and that, and then you read it too, because you cannot use something you haven’t read. So, in your bibliography, all sorts of things appear that have, as it were, passed through you. It will also be a completely different selection that you make.

Now, you have this ready-made article—it apparently happens quite often that people… well, okay, I assume you still read it, and by reading it, you still establish some connection with it. But beyond that, you haven’t really done anything, and when you see how fast something like that goes, how quickly such an article is written, it is truly extraordinarily impressive. And then it appears—well, okay, then you read it, maybe you’re satisfied with it, you change a few things here and there, but that will always feel like a downgrade because the machine, of course, does it in the best possible way—because that’s what you asked for. You wanted the best possible article.

Yes, I can’t help but make a bit of fun of it, but I actually mean it as something very, very, very serious. That as a human being, when you do it yourself and develop yourself through it, that as a human being, you give that away to a machine. I mean, if you are a musician, you could also just put on a CD, then you have the music, then you don’t have to do it yourself anymore. What is really the difference? I don’t see the difference. That you let yourself be presented with something ready-made that was created by someone else—and in this case, by a machine—that on the one hand, and on the other hand, that you conduct the entire literature research yourself and unfold your entire knowledge in a written piece through which you develop yourself. You become a different person because of it. Try it once, and you’ll notice it: when you write a piece, you are a different person than before you started.

Well, in those three minutes that the machine writes that piece for you—maybe that’s exaggerated, but in a short time—you really haven’t become a different person. I am afraid you only become less because what you should have produced with your willpower, you hand over to the machine. And again, then no one ever needs to make music again, because you can get everything ready-made.

You could also say, for example: Look, here is an example of the fugues that Johann Sebastian Bach wrote. Write me a few too? You’ll get them, you can play them, and they even resemble it. And what is the difference? What is the difference between the fugue that Bach wrote from his soul and his skill and the result of an imitation action by a machine?

I want to try, in these videos, to give examples of the absurdity of the claim that artificial intelligence could surpass human intelligence. But we must be aware that, in human intelligence, it is always about the application of will. You cannot write a paper without will. You must—and that is something different from simply entering a few things like, "Well, this is how I want it," and then it just appears. Everyone must understand that, aside from the fact that it is fraud—because if you submit such a paper, well, it is, of course, completely absurd that you do that. But if your paper—because it isn’t yours—you could also ask your better classmate, or your better fellow student, to write a piece for you because they can do it better. But at least then, it is still human, then there is still much in that piece.

But here, it is a completely will-less occurrence, and then the positive artificial intelligence specialist will probably come and say: "Yes, but the machine also has will!" But you can only say that if you have no idea what willpower actually is. Nothing happens here that has anything to do with willpower. But anyway, what I wanted to bring up in this video today is, first of all, the example of the future—the near future—of artificial intelligence in our working lives, that it could be completely taken over by machines. You really should think about that and consider how that machine currently functions—but maybe there are better versions, but even then. What should we expect?

If you want to see a doctor and you receive a message from the hospital saying: "Well, we can do a video call, or we can handle it over the phone," and it turns out that you have to communicate with a non-human? That you have to communicate with an intelligent medical system? Maybe it will eventually get to the point where the system can also look at your skin color, that there will be methods to measure your heart rate, and so on. Just imagine that happening to you! That is, after all, the wish for the future, because, they say, humans make mistakes, artificial intelligence does not, because it knows everything!

But that’s not true, is it? It’s not just about what you know, is it? It’s also about what you want! As a human being, as a doctor, you can *want* to be a healer. That is will, that goes together with knowledge, but that machine does not have will.

That was one thing, and the writing of the "best paper in the world"—on a particular subject—was the other. My appeal is also to reflect on that, to experience the difference between a human being writing a paper—which is, after all, the whole point, that they develop themselves through it—and what happens to the non-writing human who hands it over to a machine.

It is not the result that counts, it is the path toward it.

The Perfection of Artificial Intelligence by Mieke Mosmuller