AI research quality crisis: hallucinated citations and quantity over quality

A final-year undergraduate researcher, active in AI since high school, voices growing frustration with the declining quality of AI research. They point to multiple alarming trends: papers containing hallucinated citations and even leftover prompts from AI tools, misleading data that doesn't tell the full story, and labs that have built a culture of quantity over quality—pumping out papers, citing each other, and listing every lab member as co-author to inflate publication records. High schoolers are now commonly submitting to conferences through paid 'research programs' that exploit the competitive landscape. Even top labs are not immune; the recent TurboQuant controversy is cited as an example of misleading or unrepresentative work.

The user notes that genuine research from lower-tier institutions is being drowned out because it doesn't generate clicks or LinkedIn views. The irony is that AI itself, which the user loves for its creativity, has become a tool for churning out low-quality cookie-cutter work. While they still use AI for polishing writing and generating plots, they see a fundamental shift where the culture around AI research is eroding under the weight of slop. This is not just a personal rant—it reflects a broader trend that threatens the integrity and future of the field.