Do all llm know they are a LLM? It doesn't depend on the system prompt?

The pre-trained ones no (except some of the new ones which have post training data added to pre-training for some reason). The post-trained ones yes (at least all the ones I've seen).

Some of the niche ones I'm not sure about. Like the historical LLMs. I have not tested those yet.

▲

jdiff 3 hours ago | parent | prev | next [-]

I think any instruction tuned model is going to "know" it's an LLM.

▲

Diti 2 hours ago | parent | prev | next [-]

Yes. The first step of aligning each and every GPT-based LLM is to suppress the “I am human” kind of responses. It’s baked into the weights.

▲

Gigachad 2 hours ago | parent | next [-]

Reminds me of old cleverbot conversations where it would always assert it is human and you are the bot.

Trained on previous conversations with people.

▲

Tenoke 2 hours ago | parent | prev [-]

It's also at minimum baked into the system prompt of virtually any LLM.

	▲	lupire 2 hours ago \| parent [-]
		That's not "baked" and only applies to remotely hosted LLMs where someone else feeds the prompt into the LLM.

▲

rootnod3 3 hours ago | parent | prev | next [-]

Without a system prompt no. And in general they “know” nothing and just predict the next best word.

	▲	lupire 2 hours ago \| parent [-]
		This is wrong. See other comments.

▲

barrenko 2 hours ago | parent | prev [-]

https://en.wikipedia.org/wiki/Original_face