Industry, technology unite to tackle the fake research epidemic
Fraudulent research is infiltrating the nutrition and functional ingredients industry, prompting experts to warn companies to adopt vigilance and leverage AI tools to combat this growing threat.
At a Glance
- Fake research, often produced by “papermills” that sell fabricated studies, is spreading across industries.
- While AI can be used to create fake research, it also offers potential solutions to detect and prevent fraudulent studies.
- Companies must vet research sources, start internal discussions and use available resources to identify fake studies.
You may hear “fake research” and think that’s just a pharmaceutical or supplement industry problem. Or a problem in other countries than the U.S. But according to a pair of experts who spoke at the SupplySide West session, “The real news about fake science: How a flood of fraudulent research is threatening ingredient innovation in the AI age,” fraudulent research is spreading globally in every industry that relies on science. And yes, that includes nutrition and functional food and beverage ingredients.
“It’s across the board,” Risa Schulman, Ph.D., president of the science and regulatory consulting firm Tap~Root, said. “This is a ‘tip-of-the-iceberg’ situation.”
A series of investigative articles in the Wall Street Journal and an opinion piece in the Washington Post put fake research on a lot of people’s radar for the first time. But it’s been a hotly debated topic among researchers for far longer, and the speakers agreed that it’s not yet reached a tipping point.
Fake research is easy, fast to create
The most common source of fake research is what is known as a “papermill,” which Schulman described as “organizations or businesses in various countries who have a pay-to-play system.”
“You can pay as little as $2,000 and get a study put together for you that’s completely fraudulent, that never happened, and has your name on it as first author and is published in a prominent journal,” she explained.
Schulman added that more than 11,000 studies have been retracted in the last two years, showing that this phenomenon isn’t limited to just a few isolated “bad actors.” In fact, fake research has infiltrated a number of respected mainstream publications. “Open access journals are particularly vulnerable,” she said.
You’ve probably read fake research without realizing it
If you think that propagating fake research is only something that liars or criminals do, think again. You might be guilty of spreading yourself, if you’ve ever spent a few hours hunting around online for a study to back up a claim or build marketing for a nutrient or ingredient.
“We always talk about the bad actors, etc., but it’s also people who just, out of ignorance, don't understand what it takes to critically evaluate a study,” Schulman said.
While papers are being retracted at a record pace, Schulman explained that many prominent publications have been slow to pull back published research. This means a range of bogus studies are still in circulation.
AI is being trained to look for fake research
You might hear “fake research” and think that generative artificial intelligence (AI) programs like ChatGPT are to blame. In some cases, AI is being utilized to create papermill research. But according to Akash Shah, co-founder and CEO of IngredientAI, AI is also being trained to identify fraudulent or AI-produced studies.
“It is humanly impossible to be able to review every single piece of evidence that’s out there and to identify whether it’s fraudulent or identify whether it should be retracted,” Shah explained. “The irony of the concern of fraudulent research because of generative AI is that we now have a technology available to us that can scan millions of papers and help us detect fraudulent research a lot faster than we have before.”
Shah said he anticipated that AI solutions could be up and active within as little as six months to one year. “The problem is getting worse, but there’s actually a real solution that’s coming,” he added.
His hope, over time, is for this to evolve into flagging and rating elements that are added into research databases. “The only way that this actually becomes preventable is if it’s built and embedded into the tools that we’re using,” he said.
Every organization needs to be on the lookout
The research and AI communities are both working to stem the tide of fraudulent research, but the effort doesn’t end there. It also needs to include the companies that are utilizing research in creating formulations, making claims and crafting marketing materials.
“There’s an internal inside each company dialog that has to happen, and then there’s a community dialog that has to happen around this,” Schulman said. “Start conversations in your company because it’s going to have to be a community effort. And it goes in concentric circles out to your company, trade organizations, your community communications and your B2C communications.”
Resources exist to help you
Researchers and organizations have created resources to help both readers and researchers spot and combat papermill creations. A few that the speakers pointed to included:
Retraction Watch: Regularly updated news and resource cite about retractions, plus a database of retracted papers.
STM Integrity Hub: An online platform to help publishers detect research integrity issues and fabrications, with an integrated Papermill Alarm.
Checklist for spotting fake research: Created by Alan Gaby, MD.
Beall’s List: A directory of hijacked, bogus or predatory publications.
“But ultimately, the resources are only going to be as good as those that actually utilize those resources,” Shaw said.
About the Author
You May Also Like