I think "childlike" comes close but misses the mark a bit. It's not that the LLM... | Hacker News

Hacker Newsnew | past | comments | ask | show | jobs | submit

		ckrapu on May 13, 2023 \| parent \| context \| favorite \| on: Prompt injection explained, with video, slides, an... I think "childlike" comes close but misses the mark a bit. It's not that the LLMs are necessarily unintelligent or inexperienced - they're just too trusting, by design. Is there work on hardening LLMs against bad actors during the training process?

hxugufjfjf on May 13, 2023 | [–]

Yes there is. The paper describing GPT4 had a section on this.

amelius on May 14, 2023 | [–]

How about: unreliable child savant.

You can use it in sentences such as:

Would you let your children talk to the unreliable child savant?

Consider applying for YC's Winter 2026 batch! Applications are open till Nov 10
Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact