一篇在Hacker News上发表的文章对当前生成式AI模型提出尖锐批评，称其被严重过度宣传。1作者认为"前沿AI仍然只是一个高度增强的随机鹦鹉"，质疑这些系统本质上是高级的随机文本预测工具，而非真正的人工智能。1
文章指控AI公司通过大规模侵犯知识产权从事"人类历史上最大规模的盗窃"，涉及未经许可使用开源代码和数据。1作者进一步主张，"几乎来自任何大型语言模型的每一个输出都完全违法"，理由是基于许可证侵犯。1作者同时批评AI公司利用虚假的"AI威胁论"进行监管俘获和逃避责任。1
在社会影响方面，作者对未来前景表示悲观。1他预测在接下来的数十年内，白领劳动力将因自动化而大幅被替代，这一时间线与气候变化的预期轨迹相重合。1作者反对流行的"AI将杀死我们"的叙事，相反认为真正的危险源于人类社会结构的失调，而非AI技术本身。1作者在文章末尾指出这篇评论"在不到2小时内仓促完成，未借助任何大型语言模型的辅助、分析或优化"。1
An author on Hacker News has published a critical assessment of contemporary generative AI models, characterizing them as fundamentally limited technologies being promoted far beyond their actual capabilities 1. The author argues that frontier AI systems are "still just a highly augmented stochastic parrot," contending that current large language models do not represent genuine artificial intelligence but rather sophisticated text prediction mechanisms 1.
The critique extends beyond technical limitations to accusations of widespread intellectual property violations. The author claims that AI companies have engaged in "the greatest theft in all of human history" by using open-source code and data without authorization 1. Furthermore, the author asserts that "virtually every output from any of these LLMs is completely illegal" based on licensing agreement breaches 1. The author emphasizes that this commentary was "crudely written in under 2 hours without the assistance, analysis, or hardening of an LLM," suggesting the points can be made through traditional analysis 1.
The article also addresses regulatory and societal dimensions of AI deployment. The author contends that companies are employing what amounts to "AI will kill us" narratives to achieve regulatory capture and avoid accountability 1. Rather than accepting existential AI risk framings, the author identifies the genuine danger as stemming from structural problems within human society rather than from AI itself 1. The author predicts that white-collar labor will face substantial automation-driven displacement across the coming decades, a timeline that the author notes coincides with climate change trajectories 1.
评论
还没有评论，欢迎留下第一条。