I Used AI to Fake a Scientific Paper. Here’s What Happened
Science has a spam problem. And the spam filter is getting clogged up. According to all reports, academic publishing is drowning in what has come to be called AI slop: low-quality, mass-produced content generated with AI. By 2025, an estimated 57 percent of all published academic articles exhibited detectable evidence of large language model (LLM) influence, a massive jump from 12 percent in…
Science faces a significant spam problem with the rise of AI-generated academic content. By 2025, it's estimated that 57 percent of published articles may exhibit evidence of AI influence, up from 12 percent in 2023. Over 12 percent of Nature Communications reviews were found to be entirely AI-generated that same year. However, these figures likely underestimate the true extent of AI involvement, as only one in every 40 papers showing statistical evidence of AI usage formally discloses it.
The proliferation of AI-generated papers poses a serious threat to the integrity of scientific research. Amanda Sulicz, IEEE's research-integrity manager, warns that AI slop is not just damaging the scientific record, but undermining the very foundation of scientific progress that has been built over centuries. However, quantifying the scale of the problem and providing concrete examples of how easy it is to create convincing AI-generated papers remains a challenge.
To explore this issue, I attempted to create a fake scientific paper using off-the-shelf AI tools. Borrowing the concept of "vibe coding," I prompted large language models (LLMs) to generate hypotheses, code, and text with minimal human intervention. With free versions of ChatGPT and Gemini, I aimed to produce a convincing paper on optical solitons, a topic in my area of Ph.D. research.
Despite initial challenges, I managed to generate a believable paper through multiple prompts, carefully crafting each section and converting it into the required style for IEEE Access, a multidisciplinary open-access journal.
Written by urgent.news from IEEE Spectrum's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
