There's a cliff just off the highway.
https://gyazo.com/5ad8f9a0b4025aa712acdbd658ed6cc0
1: First there is a cliff, and only a few people can reach the top by climbing the cliff.
2: Efforts to make the steps smaller will gradually increase the number of people who can climb, but there is another cliff at the end of it.
3: Innovation! The cliff is gone! I was overjoyed and pushed forward, only to find more cliffs ahead.
There's a cliff just off the highway, watch out!
1: Programming difficult...
2: Libraries are easily included via pip or npm, language specifications are available on the official website with helpful tutorials translated into various languages, and there are also extensive online learning courses...
3: Burn, LLM! Libraries break daily with updates! There is no documentation on the behavior of the prompt! Is the documentation in English? No, read the source!
The LLM prompts are the equivalent of a programming language, but for the time being, there are no systematic explanations, just "I did it this way and it worked! I did it the same way, but it doesn't work! It will be "I did it this way, and it worked! The time will come when those who can empathize with LLM will play an active role, just like the oddballs who can become computer feelings played an active role in the past. Why do you make the prompt like this? It's true that it works, but it's not surprising if you think about it from the LLM's point of view. LLM natives are now saying to me, "Why do you do it like this?
relevance
---
This page is auto-translated from /nishio/高速道路を出た先に崖がある using DeepL. If you looks something interesting but the auto-translated English is not good enough to understand it, feel free to let me know at @nishio_en. I'm very happy to spread my thought to non-Japanese readers.