Ask HN: AI 感觉像是新的镭热潮，我们应该如何监管它？
1 分•作者: hamburgererror•8 个月前
感觉现在的大语言模型（LLM）就像经历了一场新的狂热，类似于19世纪末可卡因的普及[1]或20世纪的镭热[2]。这两种产品都未经管控地在药店出售，承诺能带来不切实际的效果。后来，人们意识到它们的危险性，才开始出现立法，并最终被禁止。
由于ChatGPT才诞生3年，感觉我们正处于炒作的开端。目前，我们正逐渐开始了解这项技术的负面影响：互联网上充斥着AI生成的垃圾内容，关于认知能力下降的研究，社交媒体上日益加剧的极化现象，用户报告的鼓励自残等恶劣行为。
可卡因和镭已经因为立法而从消费品中消失。很难想象如果它们今天仍在被使用，会造成多大的损害，但毫无疑问，后果将是巨大的。
我认为AI会长期存在，但也许我们应该尽快对其进行监管，以免它对社会造成过大的损害。
你们对此怎么看？
[1] https://en.wikipedia.org/wiki/History_of_cocaine#Popularization
[2] https://en.wikipedia.org/wiki/Radium_fad
查看原文
It feels like with LLMs we are living a new craze similar to the popularization of cocaine in the late 19th-century [1] or the radium fad in the 20th-century [2]. Both of these products were sold without control in drugstores under the promise of unrealistic results. Then, only when people realized how dangerous they are, legislation appeared and they were forbidden.<p>Since ChatGPT is only 3 years old, it feels like we are living the start of the hype. Currently we are slowly starting to understand the downsides of this technology: internet filled with AI slop, study on cognitive decline, increasing polarisation on social media, users reporting terrible behaviors like encouraging self-harm, etc.<p>Cocaine and radium have disappeared from consumer grade goods thanks to legislation. It's hard to imagine how much damage they would have caused if they were still in use nowadays but no wonder it would be huge.<p>I think AI is here to stay but maybe we should hurry in regulating it before it damages society too much.
What do you guys think about this?<p>[1] https://en.wikipedia.org/wiki/History_of_cocaine#Popularization<p>[2] https://en.wikipedia.org/wiki/Radium_fad