Much of the theory of intelligent design is concerned with distinguishing intended objects from random works or laws of nature. However, within the multitude of unquestionably designed objects, questions about intent may still arise. Art historians could question the authenticity of paintings by looking for evidence of tampering or copies. Teachers may suspect plagiarism in a student’s homework. And now magazine editors are dealing with very serious questions of malignant design in scientific articles. In situations that editors never faced before the computer age, they find examples of forged papers generated by artificial intelligence programs. Some are so good they even fool the peer reviewers.
Since forensics is an example of smart design in action, ID inference tools must be used to aid in this endeavor. William Dembski illustrates with ID principles within the class of designed objects in his portrayal of the case of District Officer Nicholas Caputo, who was caught listing his favorite candidates more often than happened to be possible. Today’s examples are even more difficult to crack. Often there is no element of chance or natural law that can be eliminated. The research paper, the culprit, and the AI ââsoftware are all designed intelligently. So what?
A news feature in nature speaks for the growing problem of fabricated research papers. Reporter Holly Else gives hope that there may be ways to detect malicious intent: “Strange twists and turns may indicate a foul in science,” she begins.
In April 2021, a series of strange sentences in magazine articles aroused the interest of a group of computer scientists. Researchers could not understand why researchers use the terms “fake awareness”, “deep neural organization” and “colossal information” instead of the generally accepted terms “artificial intelligence”, “deep neural network” and “big data”.
Further investigation revealed that these strange terms – which they “Tortured sentences” – are likely the result of automated translation or software that tries to disguise plagiarism. [Emphasis added.]
Since AI cannot yet mimic the cultural nuances known to human writers, an AI algorithm programmed to replace words with synonyms can become a cultural one gaffe. It is no problem to replace âcloud computingâ with âhaze figureingâ. A human editor can recognize the humor in a newspaper that speaks of âsignal to noiseâ as a âflag of turmoilâ. This lack of nuances in the software can help integrity detectives for now – until the AI ââcatches up.
Back in March in nature, warned Matthew Hutson of “Robo-Writer: The Rise and Risks of Speech Generating AI”. The latest iteration of a demonstration program called GPT stunned viewers:
In June 2020 a new and powerful artificial intelligence (AI) started dazzling technologists in Silicon Valley. It was called GPT-3 and was developed by the research firm OpenAI in San Francisco, California, and was the latest and most powerful in a series of “great language models”: AIs that Generate flowing streams of text after absorbing billions of words from books, Articles and websites. GPT-3 had been trained on around 200 billion words at an estimated cost of tens of millions of dollars.
It was fun until magazine editors started seeing GPT-3 generated papers that passed peer review. A worrying graph in Hutson’s article shows an exponential growth in artificial neural network parameters since 2018. What will GPT-n be able when its computer connections match or exceed the neural connections in the human brain? The quirky robot Data (played by human actor Brent Spiner) in Star Trek: The Next Generation, acts proactively. Data continues to strive to be a capable human imitation as it strives to understand human emotions. With more input from billions of words, GPT-n may be able to laugh at his own past missteps, such as:
AI Professor Robert J. Marks (Thought Matters) and the neuroscientist Michael Egnor (Evolution news) assure us that AI will never gain consciousness or self-awareness. This gives us the philosophical hope that data will never quite âarriveâ as a plausible human imitation. Hutson warns, however, that there is nothing stopping GPT-3 from generating misinformation, hate speech, and terrorist propaganda. Garbage in, garbage out. The advent of large language models such as GPT-3 poses two new challenges for ID theorists: staying ahead of rapid technological change (an “arms race”) and integrityof intelligence when everything that is examined is designed intelligently.
Integrity Detectives and “Tortured Phrases”
Holly Else describes how integrity detectives search for “tortured phrases” in other magazines. One particular magazine found other indications of counterfeiting:
To dig deeper, the group downloaded all of the articles published in Microprocessors and microsystems between 2018 and 2021, a period chosen by them as an updated version of GPT was released in 2019. They identified around 500 “questionable items” based on various factors. Their analysis found that papers published after February 2021 had an average adoption time five times shorter than those published before that date. A large proportion of these papers came from Authors in China. And had a subset of papers identical submission, revision and acceptance datesmost of which appeared in special editions of the magazine. That is suspicioussay the authors. In contrast to standard editions, which are supervised by the editor-in-chief, special editions are usually proposed and supervised by a guest editor and focus on a specific area of ââresearch.
But then, she continues, the detectives faced an accelerating arms race:
Microprocessors and microsystems wasn’t the only affected title – the researchers found out too Evidence of tortured phrases in articles published in hundreds of other magazines. âPreliminary research shows that several thousand papers with tortured phrases are indexed in large databases“They write, adding that” other tormented phrases related to the concepts of other scientific areas are yet to be debunked“.
The beetle swarms Starship Troopers come to mind, or the multiplying fighters in The matrix who keep coming back, no matter how hard the protagonists fight. Without better tools to quickly separate quality from quantity, the legitimacy of scientific literature is jeopardized. “It harms science,” said a detective who found 860 cases of tortured phrases in a citation database. Another suggests that the number of manufactured papers discovered so far is probably the tip of the iceberg. Jennifer Byrne from the University of Sydney commented:
âThese papers were too found because they were of very poor qualitybut it could be more plausible AI generated papers within the literature that harder to see, “She adds.
Other problems in undisputed designs
Separating the good from the bad has long been a challenge in the scientific literature. May N. Berenbaum, the new editor-in-chief at PNASHe was appalled by the number of zombie papers that are still being quoted. In her August 10 editorial, she reported examples of forged publications that are still lurking, dating back to the early days of the Royal Society. And there’s not much she can do about it. Preventing publication would amount to prior reluctance (even if it does not require omniscience). Cleaning up bad publications would bring charges of censorship. Bad papers have value; Historians need to see the bad and the good to pinpoint the bumpy road of science. Keep track of changes in subsequent editions of Darwin’s origin, for example, helps historians infer his reactions to critics. They must be kept intact and unchanged by modern editors. Tidying up the store would also risk erasing any âsleeping beautiesâ that might be out there – forgotten papers that could awaken to a new, productive life of research.
In short, the corpus of scientific literature is a mess. Manufacturing is, of course, a mess in any human endeavor: in politics, where disinformation campaigns mislead voters; in the marketing of counterfeit Rolex watches and counterfeit diamond rings; in the arts, where plagiarism robs content creators of their rights. Where are big lies and half truths not a problem? New to our digital age is the ability of AI to create realistic content and quickly change its strategies.
ID superhero required
The opposite of counterfeiting is integrity. Holly Else uses this word three times and refers to experts who sniff out examples of plagiarism, invention, and fraud. You are currently overwhelmed. Is this a field for ID advocates to take the lead? Smart design theory beautiful includes the detection of intent, whether nefarious or benign. ID researchers would most likely agree that it is better to have a flawed paper that honestly tries to argue for Darwinism than a made-up paper by a plagiarist or computer program that argues for design.
Given the above warnings from Holly Else, Matthew Hutson, May Berenbaum, and the people they cite, the word of the day is: integrity. The ID filter is great at separating design from chance and laws of nature. Now the ID community needs new rules to distinguish the integrity of online content from fakes.
A thought experiment shows how ill-equipped materialism is for this challenge. Imagine magazine editors outsourcing the task of integrity detection to AI programs. They might be successful at the level of counting instances they were programmed to, but who is watching the watchers? Let us assume that another AI program is observing the integrity detection of the AI ââprograms. The same question goes on in an infinite series.
Real flesh and blood people at some level who value integrity must be involved. A natural follow-up question follows: How has natural selection produced integrity in the human mind? If Darwinism values ââonly survival, integrity is a phantom. It is just a mind that can have a temporal value in a culture or time, but not an ontological existence. A Darwinist worldview cannot argue that integrity is inherently good – it pays to die for it – no matter what the consensus thinks. In contrast, ID proponents have a reason to value integrity: it is an eternal value that is imprinted on every human soul through conscious design. You alone can therefore tackle the challenge of separating the good from the bad in scientific content without leaving the ship when the going gets tough. Let ID advocates rise to superhero status to help resolve this growing crisis.