Dec. 2025
Last Modified: Sun Dec 14 21:45:42 UTC 2025
ハリボテ
2025-12-15 [Mon] 06:42
新山は、きらびやかな「装飾」やら「デザイン」やらをまったく理解しないらしい。
昨日某所へ行ってみてが、何が楽しいんだか全くわかんね。
そもそも人はなぜああも飾りが好きなのだろう。
つらい日常から逃避できるから? そもそも新山は逃避したい日常などないし、
むしろ日常のほうが楽しい。
Owfey, it's already the next month again.
I put it in a quote because of the randomness of the definition.
People no longer care what it actually refers to. It could be an agent,
a model, or whatever fancy skin that they wrapped a bare model.
It would be an "AI". But I always mean LLM.
Anyway, there are two problems with this "AI" thing.
- It's very unscientific. Just yesterday, I saw people arguing if
ChatGPT/Gemini has better quality than the other. The two has totally
opposite opinions. And they are not talking about moral or style
opinions. Those things can be highly subjective, but I find it crazy
that people are this divided over the pure quality of a product. I've
seen in the past that people had a similar argument over the quality
of search engines, but those were simpler times. Search enginers were
still kind of easy to evaluate. Now there are so many use cases and
variables that no single person can have a consistent opinion about
the quality of "AI".
- It's technically unsustainable. Again, I'm not talking about
their insane capital expenditure or financial bonanza. This technology
is sitting on its training data, and we have already used the most
data on earth. Yet we're barely scratching its usefulness and its
errors are very stubborn. We need another batch of earth-size "good" data,
but the supply is running out. This is exacerbated by multiplying
AI slops, and the reduction of entry-level jobs that is already happening
in software industry. We don't seem to have a long term vision.
I'm certain that the "AI" will change our lives in a very profound way, but
probably not a positive way. I'm thinking of nuclear technologies.
Nuclear weapons have changed our perception of the world, and a signle
person cannot do much about it. I feel the "AI" has similar
properties; it just increased the uncertainty and risk of our lives,
and we cannot go back. We can now kill ourselves not just physically,
but also mentally.
息子が起きてきた。
Yusuke Shinyama
Document ID: 38733000eb67bfd7f0508f8fb749d721553b59be6c242bc97fca3061367d00f3