LLM agents could potentially do the work of intelligence analysts in a fraction of the time and for a fraction of the cost, which would enable the state to aim its all-seeing eye toward anyone, not just its highest-priority targets.
文章提出了一个令人震惊的观点：大型语言模型（LLMs）可能极大地加速了大规模监控，使监控的范围从高优先级目标扩展到任何个体。
However, in a narrow set of cases, we believe AI can undermine, rather than defend, democratic values. Some uses are also simply outside the bounds of what today’s technology can safely and reliably do. Two such use cases have never been included in our contracts with the Department of War, and we believe they should not be included now:
Nature-Artikel: https://www.nature.com/articles/s41586-024-08545-z
Lewis, D. (2020). ‘We felt we had beaten it’: New Zealand’s race to eliminate the coronavirus again. Nature. https://doi.org/10.1038/d41586-020-02402-5
Edward Snowden disclosed in 2013 that the US government's Upstream program was collecting data people reading Wikipedia articles. This revelation had significant impact the self-censorship of the readers, as shown by the fact that there were substantially fewer views for articles related to terrorism and security.[12] The court case Wikimedia Foundation v. NSA has since followed.
Google's move to release location data highlights concerns around privacy. According to Mark Skilton, director of the Artificial Intelligence Innovation Network at Warwick Business School in the UK, Google's decision to use public data "raises a key conflict between the need for mass surveillance to effectively combat the spread of coronavirus and the issues of confidentiality, privacy, and consent concerning any data obtained."
the absence of a social contract
actual level of consent of individuals being documented (and by whom? by private corporations, mostly)
Patrick Ball—a data scientist and the director of research at the Human Rights Data Analysis Group—who has previously given expert testimony before war crimes tribunals, described the NSA's methods as "ridiculously optimistic" and "completely bullshit." A flaw in how the NSA trains SKYNET's machine learning algorithm to analyse cellular metadata, Ball told Ars, makes the results scientifically unsound.