Mind Matters Natural and Artificial Intelligence News and Analysis
Programming code script abstract screen of software developer.
Photo licensed via Adobe Stock

Computers Are Not Persons Because Computing Is Not Thinking

Walter Bradley Center director Robert J. Marks discusses the issues with human dignity advocate Wesley J. Smith

Recently, Walter Bradley Center director Robert J. Marks interviewed human dignity advocate Wesley J. Smith on the seeming science fiction question of “Can a computer be a person?” (November 10, 2022, podcast 212):

Here are a couple of highlights:

About Alexa:

Wesley J. Smith: I was going to ask you about Alexa because she may come on in behind me. Of course, she’s not a she. That’s just a female-sounding voice. But I can ask, we’ll call her A, so she doesn’t come on, what time it is and she’ll tell me immediately. I can tell her to play certain music and she’ll play it immediately. How does that operate? I mean, she’s not… That program is not intelligent, is it?

Robert J. Marks: No, it isn’t. We go back to something that John Searle mentioned that the reason that AI can’t understand. He said that, I don’t know Chinese, but imagine me in a room with a bunch of file cabinets and through the doors slipped a little question that’s written in Chinese. And he said, “I’ll go to these file cabinets and I’ll look until I get a match. I don’t know what it’s saying, but I’m going to look until I get a match and I’m going to copy down the answer and I’m going to slip that answer in Chinese through the door.”

The Chinese Room experiment:

Robert J. Marks: Now, from the outside, it kind of looks like whatever is inside that room knows Chinese, understands Chinese, and it’s just astonishing. But Searle in the room doesn’t understand Chinese. It’s the same thing with Alexa. When you ask Alexa a problem, it is in this humongous room, which probably includes all of Wikipedia. We have the memory and we have the computational resources to do this now. It is in this humongous room. It does some language recognition on your voice, and it goes to this big humongous room and it looks through all these file cabinets until it finds the response, which it thinks that you want.

Wesley J. Smith: It “thinks.” Is that a correct terminology there? It thinks that I want?

Robert J. Marks: Well, maybe “thinks” isn’t the right word, but I will say that Alexa screws up a lot.

Wesley J. Smith: Yes, but it’s programmed to give the response that its prior programming, its prior experience would seem to indicate is correct. Is that a good way to put it?

Robert J. Marks: Yes, yes.

What makes AI so powerful?

Robert J. Marks: It is number one, the algorithms. And we should define an algorithm since we’re going to use it.

Wesley J. Smith: That’s right. That’s my next question.

Robert J. Marks: Okay. An algorithm is a step-by-step procedure for doing something. If Google Maps, when they tell you how to get from point A to point B is giving you an algorithm. You go down I-35 for two miles, take Exit 32A, take a right at the light, et cetera, et cetera. So that’s a step-by-step procedure for doing something.

An algorithm is nothing more than a recipe. In the book, I use the example of a German chocolate cake. You have the input, which is all the ingredients in the cake, but then you have the algorithm, the step-by-step procedure that you have to do to make the cake. You put the mix and you stir in the milk and do all of this stuff. So that’s an algorithm.

Now, it turns out that the only thing computers can do are algorithmic. If something is non-algorithmic, it is not computable. And if it’s not computable, it’s something that computers can’t do. And the interesting thing is that this is true not only for today’s computers, but yesterday’s computers and computers of the future. Doesn’t matter how fast, doesn’t matter how incredible they do, they are still not going to be able to do non-computable things.

It was shown back in the 1930s by Alan Turing that there were certain problems that were non-computable. Most undergraduate computer scientists are introduced to this through something called a halting problem.

Without getting into the details Turing showed mathematically that this was non-computable.

There have since been a number of different things which have shown to be non-computable, which means it can’t be done by a computer. Now, if that’s the case, we have to ask ourself, are there things which humans do, which are non-computable? And those I would claim include sentience, consciousness, understanding, and creativity.

Dr. Marks’s book is Non-Computable You: What You Do That Artificial Intelligence Never Will (Discovery Institute Press, 2022). An excerpts is available here (Chapter 2).

Additional Resources

Podcast Transcript Download

Mind Matters News

Breaking and noteworthy news from the exciting world of natural and artificial intelligence at MindMatters.ai.

Computers Are Not Persons Because Computing Is Not Thinking