Nicholas Polson has authored 258 academic papers in 2026 so far
An economics and statistics professor has posted 258 AI-generated preprints in 2026, prompting questions about academic integrity, mental health, and the “publish or perish” incentives that reward sheer volume over quality. Commenters debate whether large language models should be used to draft or even review research papers, warning of a future flooded with convincing but shallow “slop” that strains peer review and erodes trust in written scholarship. Others note that many of these manuscripts resemble internal notes more than serious journal articles, highlighting how preprint platforms and citation metrics can be gamed in an AI-saturated research ecosystem.
AI-generated papers and academic incentives
- Many assume the bulk of the 258 “papers” are LLM-generated or LLM-heavy.
- Commenters argue this behavior is strongly incentivized: more submissions bring more money, prestige, and grants for individuals and institutions.
- Some see it as exposing how meaningless “paper count” has become; others fear it further erodes standards and wastes reviewers’ time.
- There is debate whether this is a deliberate stunt, opportunistic careerism, or something more personal; motives are viewed as unclear.
AI in reviewing, code, and workflows
- Several note that AI is already used for peer review and code review, often unofficially.
- Supporters think AI reviewers could catch factual or methodological errors.
- Skeptics predict over-reliance, lazier human review, and compounding errors as flawed AI-reviewed work is used to train future models.
- In software, some describe near-closed loops: AI writes code, AI reviews it, AI tests it, and sometimes AI “uses” the app.
Writing, “slop,” and the future of text
- Multiple commenters lament a flood of low-quality, LLM-style prose (“slop”), including in corporate documents and slide decks.
- Some predict written media will lose value and in-person speech will regain status; others find this prospect disturbing and argue writing is foundational to civilization and long-term knowledge.
- There is anxiety that people are losing deep reading and writing skills, accelerated by AI but also by video platforms.
Quality, correctness, and detection
- Experiences with LLM-written drafts suggest current models still produce weak or error-prone academic text without heavy human editing.
- Some imagine AI “journal clubs” that systematically check papers for concrete issues (e.g., data/image reuse, missing controls) rather than judging truth.
- Tools that claim to detect AI writing are viewed as imperfect and sometimes overconfident.
Ethics, reputation, and mental health
- Some ask why an established academic would risk reputation on mass-generated work; others counter that academic reputations are surprisingly resilient.
- A few raise the possibility of AI addiction or AI-amplified mania, where rapid, effortless generation of “papers” becomes self-reinforcing.
- There is concern that such behavior, whatever the cause, normalizes dumping internal notes as “papers” on preprint servers and further confuses what counts as serious research.