Review articles are getting bland and AI-indistinguishable, scientist warns
NikoMcCarty · x · 2026-09-29
SynBio researcher Niko McCarty argues that modern review articles have become voiceless literature lists that "could easily be written by AIs" — and often are. He contrasts them with classic opinionated reviews: Carl Woese's 1987 attack on the prokaryote-eukaryote dogma, and Doolittle & Sapienza's 1980 selfish-gene review, which made a compelling case with only 54 references, versus today's norms that often mandate 120+.
He blames rigid journal rules on length, figures, and citation counts, plus reviewers who take offense when their work isn't cited, and calls for looser models of scientific writing — even a journal that publishes biological ideas of any length and reference count.
His sharpest point: opinionated, free-flowing writing may be the last protection against AI, since models can scrape and summarize papers far faster than humans. Science should incentivize hot takes and spiciness, which requires editors with taste willing to overrule negative reviewers.
More from AGI Musings
- Gladstone AI interviews a dozen diplomats: no US-China AI treaty, just a crisis phone call — harris_edouard · 2026-09-29
- Ben Thompson: pre-built UI is dead, on-demand agent-built apps are the new era — rseroter · 2026-09-29
- Before AI slop there was human slop: React bloat, stock photos and template sites — vhanagwal · 2026-09-29
- With millennium-prize-tier model swarms, should AI safety conjectures be written down? — inductionheads · 2026-09-29
- So8res: antitrust law may bar AI companies from coordinating to slow down — wfithian · 2026-09-29
- Ramez Naam: The Intelligence Explosion Is Fighting Diminishing Returns, Not Happening — natolambert · 2026-09-29