How do we continue from here?

热度趋势

趋势数据积累中

百分比基于当前可用热度信号，而非评论数或独立用户人数。

AI 摘要

一位Reddit用户对AI的持续发展表示担忧，认为在数据中心和芯片上的巨额资本支出可能导致金融泡沫破裂。他们观察到，AI主要使现有产品更便宜或更好，而非创造大量新需求或提高产品价格。用户质疑这种“低调”的AI普及如何避免市场崩溃，特别是如果公司无法避免为可能未充分利用的计算资源付费。这与Ed Zitron等人物对潜在负面结果的担忧不谋而合。

Hey,

I'm not a hater or a fanboy, I'm just an average person using AI where using AI makes sense and being annoyed by AI being stuffed into things where it doesn't belong. I'm sure AI will transform the world, but I'm also pretty sure not next year and not completely.

What I'm really wondering, though, is how AI can become "stable" without a massive financial bubble bursting. All the AI we're using today is running on existing hardware. The spending for new data centers and chips all comes in top of the current capacity and already, the capex is absolutely insane. I know that people like Ed Zitron say that this will end really bad, but is there a numbers based study or article where somebody lays out how it all could work out without everybody being replaced by AI? Currently, what I see, is that AI is just making stuff cheaper or better, but not really creating massive additional demand. Nobody spends twice as much on Photoshop because of AI, nobody pays more for a fridge because it can say hi in the morning. Rather, people expect the AI to be a standard enhancement of the product at a similar price - just like more RAM or a newer processor used to be expected in new PCs.

Is there any meaningful way how such a "low key" AI adoption could not crash the market? Is there some way for the companies to wiggle out of paying for the compute if they don't need it?