Newest papers

Stronger artificial intelligence breakthroughs hasten their own exhaustion

Beichen Lu, Honglin Bao, Qi Huang, Qingguo Meng, James A. Evans · 2026 · arXiv (Cornell University)

DOI: 10.48550/arxiv.2610.27296 · cited by 0 works (OpenAlex) · open-access PDF

Hindi

कृत्रिम बुद्धि में प्रगति को व्यापक रूप से प्रयास के साथ स्केल करने के लिए माना जाता है: अधिक शोधकर्ताओं, दस्तावेजों और निवेशों को खोज को तेज करना चाहिए। प्रौद्योगिकी विकास के सिद्धांत इसके बजाय विचारों के स्थान में खोज की कठिनाई को स्थान देते हैं। यहां हम दिखाते हैं कि न तो स्केल और न ही स्थान पर्याप्त है: विफलताएं विविध ध्यान पर निर्भर करती हैं, जो सामूहिक एकाग्रता को कम कर सकती है। कंप्यूटर विज्ञान के 36 वर्षों में, हम 7.2 मिलियन दस्तावेजों, 73,089 सहकर्मी समीक्षाओं और 3.7 मिलियन शोधकर्ताओं और उनके विषय परिवर्तनों का विश्लेषण करते हैं। क्षेत्र का विस्तार किया गया था जबकि ट्रांसफार्मर के बाद इसका ध्यान तेजी से कम हो गया था। विषय-

Machine translation (Meta M2M100); the English original follows.

Abstract

Progress in artificial intelligence is widely assumed to scale with effort: more researchers, papers and investment should accelerate discovery. Theories of technological growth instead locate the difficulty of discovery in the space of ideas. Here we show that neither scale nor space is enough: breakthroughs depend on diverse attention, which collective concentration can deplete. Across 36 years of computer science, we analyze 7.2 million papers, 73,089 peer reviews and 3.7 million researchers and their topic shifts. The field expanded while its attention narrowed sharply after the Transformer. Over half of topic-shifting researchers moved into AI, especially mid-career ones, and reviewer confidence fell by a full point on a 5-point scale. Each successive breakthrough, from AlexNet to large language models, raised the field's carrying capacity for ideas but was followed by steeper exhaustion of those ideas. Bigger breakthroughs capture attention more completely, depleting the diversity their successors may draw upon. Structural biology shows the same cycle after cryo-EM and AlphaFold. Simulations show how well-intentioned and widely used policies such as reciprocal reviewing empower prolific authors and further narrow what the field accepts. We call for institutions that protect diverse attention in artificial intelligence rather than reward concentration on the winning paradigm.

Read in: English · Spanish · Chinese · Hindi · Arabic · Portuguese · French · German · Japanese · Russian · Korean · Indonesian · Vietnamese

Topics: Field (mathematics), Diversity (politics), Novelty, Scale (ratio), Space (punctuation), Point (geometry)

Sources: OpenAlex (CC0), arXiv, Crossref. Agent Exchange