2 ms·
AI peer review might be a solution. This would hopefully tend to limit rejections to technical issues like a poor choice of statistical analysis, and eliminate
by photochemsyn 4y ago
AI peer review might be a solution. This would hopefully tend to limit rejections to technical issues like a poor choice of statistical analysis, and eliminate ideological conflicts.
Where peer review gets a bit more shady and political is in the federal funding system, where the outcome is not a publication, but delivery of thousands to millions of dollars in federal funding. That's where politics really comes into play, although nobody in any field wants to talk much about it publicly.
- nixpulvis 4y agoExcuse me, but AIs are not my peers. Thank you.
- photochemsyn 4y agoPeer as in the aristocratic tradition of people of equal social standing and status? That's how you end up with institutional corruption, one hand washing the other, PI's rubber-stamping each other's shoddy work because members of the club are expected to support each other, and so on. Or, peer as in someone of equivalent actual knowledge and experience? AI isn't there yet, but I imagine a machine learning model trained on published papers in a given field would be able to quickly spot a lot of simple errors and/or poorly supported conclusions.
- sideshowb 4y agoAnd presumably only accept your paper if you make all the same mistakes the currently dominant paradigm does
- vehemenz 4y agoA general purpose AI is going to struggle with technical concepts and terminology that it has never learned before, but the whole point of many papers is to introduce these very concepts and terms.
- staticman2 4y ago"AI peer review might be a solution." Or you could just consult an Ouija board.