# LLMs are moving from generating artifacts to creating hyper-custom worlds on demand, but still lack the ability to natively perceive and audit what they create (Andrej Karpathy/@karpathy) — retained record

> Retained techmeme record from techmeme.

Canonical: https://highsignal.app/data/records/e4e6c02bd4abf29b

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


source: [techmeme](https://highsignal.app/data/techmeme) published 2026-08-02 21:25 ingested 2026-08-03 09:04

[original source ↗](https://x.com/karpathy)

## retained text



```
Techmeme permalink: https://www.techmeme.com/260802/p10#a260802p10

<p><a href="https://www.techmeme.com/260802/p10#a260802p10" title="Techmeme permalink"><img height="12" src="http://www.techmeme.com/img/pml.png" style="border: none; padding: 0; margin: 0;" width="11" /></a> Andrej Karpathy / <a href="https://x.com/karpathy">@karpathy</a>:<br />
<span style="font-size: 1.3em;"><b><a href="https://x.com/karpathy/status/2083749667410727319">LLMs are moving from generating artifacts to creating hyper-custom worlds on demand, but still lack the ability to natively perceive and audit what they create</a></b></span>&nbsp; &mdash;&nbsp; We're starting to leave the territory where you'd test an LLM by e.g. &ldquo;create an svg of pelican on a bicycle&rdquo;.</p>
```



document fetched 2026-08-03 09:04
