Years ago, Evolution News did a series on Intelligent Design in Action. One entry in the series involved forensics; another involved outing liars. From recent news, here are some updates about forensic science in action, and how ID is well suited for the pursuit of truth.
Unmasking Fake Art
Unless they are disclosed by the artist as such, reconstructions of famous art masterworks pose a serious problem for art dealers. While someone might buy a reconstruction, nobody wants to pay a lot of money for a fake. In a Research Highlight Nature tells how AI (artificial intelligence) is becoming a tool for art critics. “A neural network can use the surface roughness of a painting to tell one person’s brushstrokes from another’s.” Can “surface roughness” be an indicator of ID? Sure.
Kenneth Singer at Case Western Reserve University trained an AI neural network on a million photos, and then on the brush strokes of four renditions of a flower by art students.
When the system examined other paintings by the same four artists, it could attribute them to the correct painter with 95% accuracy, on the basis of scans just one centimetre square. The neural net seems to have recognized the authors from brushstrokes roughly the diameter of a single paintbrush bristle. [Emphasis added.]
Their paper in Heritage Science1 describes how the team did it. Part of the goal was to distinguish true positives (TP, correct scores) from false positives (FP, missing a fake) and false negatives (FN, missing a true). The new system had success at even microscopic scales.
The surface data were divided into virtual patches and used to train an ensemble of convolutional neural networks (CNNs) for attribution. Over a range of square patch sizes from 0.5 to 60 mm, the resulting attribution was found to be 60–96% accurate, and, when comparing regions of different color, was nearly twice as accurate as CNNs using color images of the paintings. Remarkably, short length scales, even as small as a bristle diameter, were the key to reliably distinguishing among artists. These results show promise for real-world attribution, particularly in the case of workshop practice.
How does intelligent design enter this situation? Four ways. First, intelligent designers had to write the software for the AI system and design its purpose. Second, designers had to set the goals of the software by setting targets for it to study, such as differences in brush strokes. Third, designers had to evaluate the results and score the success of the system. Last, designers had to understand how minds work, including what might motivate a forger and what methods he might use to get past art critics.
Unmasking Fake Research
The rise of “predatory journals” has given headaches to science editors. Predatory journals try to imitate serious science by using underhanded means and shortcuts. Some mainstream journals, though, have fallen prey to forgers and even to AI systems that can write fake papers good enough to pass peer review. See “Detecting Malicious Intent in Undisputed Design,” in which we described how AI can detect “tortured phrases” that stand out in the text as indicators of computer-generated papers by AI that is not yet smart enough to understand the nuances of English usage.
Nature announces another headache: “Predatory publishers’ latest scam: bootlegged and rebranded papers.” This one will be tougher to solve because there are fewer indicators like tortured phrases. Scammers are motivated to take this route, too: why reinvent the wheel? If a paper has already passed peer review and has sound science, why not resubmit it to a journal with a few modifications and gain the prestige of citation? The journal gets the money without the hard work. This problem has been growing like the mythical Hydra: punish one fake journal, and another pops up.
To solve this kind of fraud, Nature gives four policy rules to “starve the Hydra” and root out the motivations for rebranding papers. Once again, this takes intelligent design. Malicious intelligent design commits the fraud, and honest intelligent design must implement policies and methods for filtering out the fake and preserving the true.
Unmasking Fake Life
This last item is different from the other two, because it seeks to find the truth about life detection on Mars. Unlike SETI (another instance of ID in Action, to the chagrin of the evolutionists involved), this situation involves microbial life. Evolutionists will deny that microbes are products of a designing intelligence, but they still need the right tools to tell the true from the false.
Writing for Live Science, Harry Baker discusses “false fossils” that may litter the Martian landscape. NASA is eager to determine if microbial life ever existed on Mars. That’s long been one of their top priorities. Unfortunately for astrobiologists, there are numerous ways that natural chemical and geological processes can fool them. Baker reminds his readers of the 1996 rush to judgment when NASA scientists eagerly announced to the world that fossil microbes had possibly been discovered in a Martian meteorite. That incident was so influential, it motivated NASA to launch a new “science” of astrobiology (which, to date, has no “bio” in it).
Their discovery was hailed as the first proof of alien life and even prompted a speech from President Bill Clinton. However, further tests revealed that these fossils were completely abiotic, meaning they were not made by life-forms.
The article goes on to describe “a wide diversity of potential false biosignatures on Mars” that will have to be discriminated by some sort of design filter. Even though evolutionists will deny that Martian microbes were designed, they will still need to distinguish false positives. Baker shows a photo of crystals of carbon and sulfide, about the same size as bacteria, that would easily fool a layperson.
Another example are pseudo-microbialites, which mimic physical structures created by microbes, such as stromatolites — which are large structures left behind by photosynthetic algae that grow upward as cones, domes and columns. Such structures could be left behind from marine life in Mars’ past oceans, but near-identical structures can also form naturally without any microbes so it will be hard to tell if they are genuine.
There’s a key ID word: genuine. It takes a mind to discern what is genuine, and to care about authenticity. It also takes a mind with ingenuity to design a truth identifying mechanism that can filter out the genuine from false positives.
Animals can discern reality, but not authenticity. To see why, imagine a packrat scurrying around two copies of a journal: one authentic, and one fake. The papers are “real,” but the animal will readily tear bits and pieces of either one for its nest. A computer doesn’t care either. If it runs an AI system, it must borrow its concern for authenticity from the mind of a human being that can train its detectors on what to look for.
ID scientists are available to help. Their business is applying principles of discernment to certify the genuine in all kinds of cases where the truth must be told.
- F. Ji et al., Discerning the painter’s hand: machine learning on surface topography. Heritage Science volume 9, Article number: 152 (12 Nov 2021). Open access.