# Some people seems to not get this: LLMs do not need to be the ASI themselves — retained record

> Retained reddit record from reddit:singularity.

Canonical: https://highsignal.app/data/records/37c41b0cb6a073f6

[← sources](https://highsignal.app/data)


source: [reddit](https://highsignal.app/data/reddit) (reddit:singularity)  published 2026-09-23 07:16 ingested 2026-09-24 02:38

[original source ↗](https://www.reddit.com/r/singularity/comments/1wnyyy9/some_people_seems_to_not_get_this_llms_do_not/)

## retained text



```
Some people are skeptics about ASI because "LLMs can't reach ASI". Let's make this clear.  
  
If any AI-LLM based system can research and develop another AI model because it has some narrow super-intelligence on that topic, then a new model they will create might not be an LLM anymore, it could be something **completely new**.

We don't need to create that "completely new" model-architecture by ourselves for an ASI to arrive. A system more capable than us can do it, just like how it is more capable than us to solve some difficult math equations (narrow super-intelligence).

So either you consider LLMs as capable of reaching ASI or not, it doesn't matter.


```



document fetched 2026-09-24 00:33 · canonical https://reddit.com/r/singularity/comments/1wnyyy9/some_people_seems_to_not_get_this_llms_do_not
