Using AI to Discover Intelligent Design

[00:00:12] Andrew McDiarmid: Welcome to ID the Future. I’m your host, Andrew McDiarmid. Well, today I welcome to the podcast David Coppage to discuss a recent example of intelligent design in action. Scientists using AI to detect evidence of intelligent design.

David Coppage is a freelance science reporter in Southern California. He worked at NASA’s Jet Propulsion Laboratory for 14 years on the Cassini mission to Saturn until he was ousted in 2011 for sharing material on intelligent design, a discriminatory action that led to a nationally publicized court trial in 2012. Discovery Institute supported his case, but a lone judge ruled against him without explanation. He has been a board member of Illustra Media since its founding and serves as their science consultant. A nature photographer, outdoorsman, and musician, David holds B.S. degrees in science education and in physics and gives presentations on ID and other scientific subjects. David, welcome to the show.

[00:01:13] David Coppedge: Good to be here.

[00:01:14] Andrew McDiarmid: So, music. Do you play an instrument or do you sing?

[00:01:19] David Coppedge: I play the French horn and I also write compositions for orchestra and band. Oh, nice home system. Yeah, it’s a lot of fun.

[00:01:28] Andrew McDiarmid: And have you had a chance in recent times to hear a band play your compositions?

[00:01:35] David Coppedge: Yes, actually, I wrote a March for the 50th anniversary of Apollo 11, and it was premiered by the US Air Force Band.

[00:01:44] Andrew McDiarmid: Wow, that’s awesome. I’d love to hear that. Well, we’re not talking about music today, but I was curious about that. Now. You write regularly at our flagship news and commentary site, evolutionnews.org you write about plants and animals that exhibit evidence of design and irreducible complexity. You also occasionally highlight stories of intelligent design in action, from pattern matching in archaeology to the uncovering of ancient hidden cities with the forest penetrating technology of Lidar in the Amazon rainforest. You wrote recently of another example of intelligent design in action that I wanted to discuss with you today. Scientists who are using AI to analyze photographs of the Nazca Plain in Peru to find possible sites of geoglyphs. Now, tell us, what are geoglyphs?

[00:02:32] David Coppedge: Geoglyphs are patterns that were made by human beings, of which the Nazca drawings are one of the most famous examples. But in other words, they’re earthworks. They could be mounds of dirt or burial mounds or pyramids or any kind of thing. Geological glyphs. In other words, writings or symbols that were created by human beings.

[00:02:55] Andrew McDiarmid: Okay, got it. Using elements from the earth, like stone, gravel, and the earth itself. Well, before we get to the details of this pretty cool story, I want to give listeners a little review of the basics of Design detection. Now, back in 1998, mathematician William Dembski published his landmark book the Design Inference, wherein he presented a rigorous method of detecting design in nature. The design inference uncovers intelligent causes by isolating the key trademark of intelligent causes, specified events of small probability.

Just about anything that happens is highly improbable. But when a highly improbable event is also specified, in other words, it conforms to an independently given pattern, undirected natural processes or causes lose their explanatory power.

So design inferences can be found in a range of scientific pursuits, from forensic science to research into the origins of life to the whole search for extraterrestrial intelligence. Now, David, you demonstrate a firm grasp of design detection. How would you explain the importance of Dembski’s explanatory filter for detecting design?

[00:04:08] David Coppedge: Well, firm grasp might be in scare quotes or something. I struggled through the design inference 2.0 that came out a couple of years ago with Winston Ewart and William Dembsky. A lot of it was very intuitive, the first part, and he uses a lot of good analogies that help. But when he gets into the Bayesian inference and symbolic logic and set theory and equations and stuff, I struggled through it, but I got the gist of it. So anyway, the importance of the design inference is that it adds rigor to our intuitions. In other words, we all do design detection all the time. You step outside, you see water on the sidewalk and you, you’re asking, did it rain last night or did a water truck come by and spray my sidewalk? You know, so the first would be a natural cause, the second would be an intelligent cause, maybe not super intelligent, but in other words, it was intentional. Somebody did it on purpose. And so what the design inference tries to do is to separate natural causes and chance from intentional causes. And that’s what Dempski’s original design filter was designed to do, was to give you a flowchart so that you could separate out what is natural from what was intentional.

And it’s important to stress that the design inference is really only asking one simple question.

Was something designed or not designed? Okay, you got those two choices. It’s like being at the top of a ridgeline and seeing which way a marble is going to roll to the east or to the west. And, and naturally, once it starts rolling, there’s a lot of follow up questions that ensue that are very important, like, well, who rolled this marble and why did he roll this marble? You know, but those are beyond the actual design inference itself. It’s just trying to answer the question. Was something natural or chance, or was it intentionally designed?

[00:06:07] Andrew McDiarmid: Yeah, that’s a great way to put it. And I’m glad you mentioned that. This is. This is something we do all the time. You know, when we walk out of our house, even in our house, everywhere. You know, Doug X calls it the design intuition, that we are all born with this innate ability to recognize the handiwork of an intelligent agent.

Being intelligent agents ourselves.

[00:06:32] David Coppedge: Yes. And that’s what makes detective stories so engaging. And. And you know, who done it? Was it a murder or did the guy just die of a heart attack or what? And so we do this kind of thing all the time. I like to go out with my drone and take nature photographs. And I found this one place that was. This relates to our story about the Nazca lines, as I’ll illustrate. But I found this area out in the desert that has beautiful artwork of trees on the ground. I mean, it looks like artwork. They’re just beautiful designs, and you can’t really see them unless you get up high with a drone looking down at them. And I call them dendrites, but actually they’re perfectly natural. They’re drainage patterns from rainwater on a shallow slope. And it starts carving these ditches that get bigger and bigger, and. And they form these beautiful fractal patterns that look like works of art, but those can be explained naturally. And so the design inference always gives preference to natural or chance causes by default. And you have to prove that something was intelligently designed. And so that’s why the design inference is robust against false positives. Now, a false positive would be saying that something was designed when it really wasn’t. Okay. Like looking at those tree patterns and saying, oh, somebody did that. The design inference is robust against that. And Dembski explains this in his book no Free Lunch, which is another good book on the design inference, but it can make false negatives. There’s a statue or a sculpture, rather, downtown at the Museum of Modern Art in Los Angeles that looks like somebody just took a whole bunch of airplane parts and welded them all together. It looks like it’s just a big pile of junk, but it’s on a pedestal, and it’s art. You know, somebody might look at that and say, that was not designed, but it was. So that would be a false negative. And, yeah, the design interference can make mistakes like that because it gives preference to saying something is not designed unless it’s rigorously shown to be.

[00:08:43] Andrew McDiarmid: Yeah, very good point. We can see things that arose Naturally, and consider them works of art. And we can also look at works of art, especially nowadays, and think maybe they were just thrown together by some natural process.

[00:08:57] David Coppedge: Yeah. Somebody might have just poured paint on a canvas and drove over it with their car and called that art.

But you look at the difference between Mount Rushmore, which has. Clearly, we know the history of that, but if you didn’t know, you’d be able to tell the difference between that and the Old man of the Mountain, Just a shape on a mountainside that looks like a face but could occur naturally. And so that’s what the design inference does, is try to separate intentional from natural causes.

[00:09:28] Andrew McDiarmid: Right. Yeah. And I’m very glad that we have this explanatory filter. This. This ability to do that. Well, let’s talk about scientists who are using artificial intelligence to make design inferences. You call it Intelligent Design squared, Intelligent agents designing machine intelligence that is capable of detecting intelligent design. The Nazca Pampa is designated a World Heritage Site by UNESCO because of its immense geoglyphs averaging 90 meters in length. Can you describe this location in more detail? And for those not familiar with what geoglyphs look like, can you describe them?

[00:10:06] David Coppedge: Yes. I have not been to this area. I only know about it by what I read and the pictures I see. I think a lot of people have heard about it because it’s called by the World Heritage Site. UNESCO calls it, you know, the best example of geoglyphs in the world. And These things are 100 yards long, many of them. And there’s. There’s actually two types. There’s the kind that have been carved into the sand, you know, by digging ditches or whatever. And then there’s. So those are the line type and the relief type, or when they’ve lined up stones on top of the ground. And so those are two classifications of these things. And I want to hasten to add, I’m not an expert on the Nazca culture. I don’t know anything about why they made these things.

They were said to have been made starting around 100 BC all the way possibly into the Middle Ages of Europe.

So possibly a thousand years, people in this area were making these works of art. We don’t know why they did it or what they were trying to communicate, but we can tell that they were designed.

[00:11:16] Andrew McDiarmid: Right. And there is debate about why these geoglyphs were made or what they’re trying to communicate. But thankfully, we don’t need to worry about the why with the design filter. We use it to see if chance and Natural law should be considered or ruled out and confirm whether something is the product of a designing intelligence. Now, they’ve been looking for geoglyphs for about a century now. And how has the technology advanced over the years?

[00:11:41] David Coppedge: Well, at first you had to either get up on a mountain to look down and see these things, or else just walk around and notice that, wow, this is an unusual row of stones or an unusual line that seems to be drawing an animal or something. And they started finding these early in the 20th century, and they would find maybe a new one a year, one or two a year. But then with aircraft and the ability to fly and look down over them, they began to find more. And now there’s been another quantum leap of discovery of these geoglyphs with the use of AI, because with AI, they can analyze the patterns that aircraft see and the maps and see things that are very faint to see with the human eye. But AI can, can determine that, yes, something was designed and not natural.

[00:12:39] Andrew McDiarmid: Right? Yeah. AI is coming into our lives in lots of different ways these days. And as you know in your article, AI can be a good tool. It just shouldn’t be doing all the thinking for us. Tell us what the researchers did to set up and train their AI model.

[00:12:55] David Coppedge: Well, they first mapped the whole area, the whole 290 square miles where these geoglyphs exist. And, and then they cut it all up into, into segments 10 centimeters in size. And they trained their AI model on, first of all, they trained it on known geoglyphs so that the model would know what a geoglyph looks like. And then they trained it on negative models which are not designed. So they’re giving it the ability to detect intentional lines versus just natural lines. And so they trained this and came up with 1309 hotspots, in other words, spots that the AI model thought could be geoglyphs. And so then the scientists ranked These in ranks 1, 2, or 3 based on how probable they were actual geoglyphs. And they ended up with about 303 new geoglyphs that were not known previously.

[00:13:59] Andrew McDiarmid: Wow. Yeah. It’s a good example of how the scientists have to do the lag work first to train the model and, you know, take all the pictures to feed the model and then to sift through what the model provides as, hey, you might want to look at this and proceed from there.

[00:14:16] David Coppedge: And yeah, and to show the human input into this, they had to double check what the AI came up with. So they had all these hotspots, but they had to go out there on Foot and fly drones and they actually spent like 2600 extra hours of work analyzing what the AI model did. So that’s why I call it the design inference squared because it took human inference plus machine inference. And the humans of course, designed the machines that did the inference. So it’s like a cubicle arrangement of intelligence, design inference. But yeah, they had to do a lot of follow up work to verify and they found some that the AIM model said that were false positives, but then they found some additional false negatives or positives that were designed but the machine didn’t notice them. So they only found, they say, the low hanging fruits in this attempt. So they, they believe there could be thousands more out there.

[00:15:17] Andrew McDiarmid: Wow. Yeah. I was going to ask you, is there more work to be done to create a complete map of this area of the world and all the geoglyphs?

[00:15:26] David Coppedge: Oh, yes. And they actually think that what they’ve done with AI is going to be useful for archaeologists all over. We already know that LIDAR has been able to fly over Amazonia and find huge geoglyphs out there, like whole villages and cities that were unknown. I mean, the old story was that, you know, this is a native area where you have these natives running around with loincloths and just hunting monkeys and stuff, but actually they’re finding evidence of civilization out there that was unknown prior to the use of LiDAR. This has also happened in Cambodia with the temples hidden under the jungles and stuff. So, yeah, archaeologists are going to have a field day using aircraft drones and AI to determine evidences of human design.

[00:16:15] Andrew McDiarmid: Yeah, that is fascinating. And those listening to our podcast can actually go read about your coverage of what they’re finding in the Amazon rainforests. That’s@ovolutionnews.org as well. Well, do you think AI can be used constructively to further intelligent design projects and research?

[00:16:34] David Coppedge: Well, this is a very controversial question. And you know, I listened to Elon Musk talking about AI and the dangers of it. And yeah, it’s, it’s got to be truth seeking, he, he stressed. And, and if you get AI that gets good at lying, I mean, who knows what can happen? So yeah, it’s only a tool, it’s. And another thing about the design inference is that it is agnostic to morality. In other words, we’re not trying to detect whether a design was a good design or a helpful design, but just whether a design exists. And so, you know, the ethics of a design has to be left up to other fields like ethicists and, you know, Moral philosophers and theologians perhaps.

[00:17:20] Andrew McDiarmid: Yeah, yeah. Now we’ll encourage listeners to go check out some of these geoglyphs in this area themselves. But what sort of pictures and imagery were captured in these, these fascinating monuments?

[00:17:35] David Coppedge: Well, they’re very puzzling. The large line geoglyphs tend to be animals and they include things like birds and cats, snakes, monkeys, foxes, killer whales and fish. But then the smaller ones that are, that tend to be the relief type with rocks lined up in images, they tend to be smaller and they tend to have a lot of things like human motifs and decapitated heads. And one that I thought was confusing was the knife wielding killer whale. So yeah, there are some puzzles to solve about what these people were trying to communicate. Maybe they were just having fun. Maybe this was their hobby of making art, or maybe it had a religious purpose, we don’t know. But that’s beyond the capabilities of the design inference itself.

[00:18:24] Andrew McDiarmid: Yeah, well, you do make an interesting point. Also, at the end of your article you note that ID is equally valid at detecting evil designs and good designs. You know, doesn’t discriminate as to why or how or who. What did you mean exactly by that? And can you give us another example perhaps?

[00:18:44] David Coppedge: Well, the clearest one is you’re a forensic scientist trying to determine whether somebody died of natural causes or was murdered. Let’s say that a lidar specialist flies over a region and finds evidence of a mass grave. I mean, that would be an evil design, but it would be unnatural as opposed to just some natural forming fissure or trench or something. And you could look at a crater and determine whether it was a bomb crater or whether a meteorite struck the earth. So the latter would be a natural cause. The form would be an intelligent cause, although a malevolent. Malevolent intelligent cause. So again, you know, the business of intelligent design is just to define whether something was designed or not. But the ethics of these things are left up to other fields.

[00:19:36] Andrew McDiarmid: Yeah. Well, if you are interested listeners in learning more about the design inference and how to apply it in your own life or field. As David mentioned, a second edition of the Design Inference was recently published, completely revised and expanded and that came out in 2023. So you can look for that@discovery.org or anywhere you can find books, Definitely a resource you’re going to want to turn to. Well, David, this is actually the first time we’ve talked on ID the Future. And it’s a pleasure to unpack some of your writing and your thoughts on this topic. And I look forward to the next time, we’ll do it.

[00:20:13] David Coppedge: Great. Good to be with you. Thank you, Andrew.

[00:20:17] Andrew McDiarmid: Listeners, read more of David’s work@ovolutionnews.org and to get straight to his work from the evolutionnews.org homepage, click the Writers tab near the top and click on David’s image and name. That’s evolutionnews.org a wonderful, wonderful resource. Lots and lots of articles on all the topics related to evolution and intelligent design you’re going to want. Bookmark that and visit regularly. Well, for ID the Future, I’m Andrew McDiarmid. Thanks for listening.