高效估算数据表最大重叠大小，基于对齐引导方法，SIGMOD 2027录用论文，值得数据库研究者细读。
arXiv:2607.03049v1 Announce Type: new Abstract: Fast estimation of the size of the largest overlap between tables enables blocking and query-by-table …
直击大模型安全对齐痛点，PolicyAlign提出直接策略优化法，让模型既安全又少降性能。
arXiv:2606.25442v1 Announce Type: new Abstract: Safety alignment of large language models (LLMs) typically depends on high-quality supervision data, s…
OpenAI发文论证，长期AI安全研究亟需社会科学家参与，以解决人类心理、偏见与理性不确定性，促进ML与社科跨界协作。
We’ve written a paper arguing that long-term AI safety research needs social scientists to ensure AI alignment algorithms succeed when actual humans a…