A really interesting advancement in AI tool usage is with new interaction patterns.
Humans are training LLMs, but LLMs can then be used to train humans. I've even heard stories of humans being trained by LLMs on how to more effectively prompt LLMs.
And of course the final frontier is when LLMs prompt other LLMs to maximally leverage the strengths of different models.
There is an issue with the LLMs training one another and no one knows what will happen. Please don’t forget that. 🫂🫂🫂
I always use LLMs to properly prompt my request.
Promptception?
I love AI !
This is going to get very interesting very fast and I'm not sure my little human brain can handle it.
I’m using AI to assist me, in my AI programming journey Anon. Specifically ChatGPT with its voice prompt. It’s brilliant👏…
Interesting .. and potentially terrifying .. time will tell
potential feedback loop of stupidity but yeah.
Ai is great time saver but boy I worry about the kids depending on it rather than any critical thinking for themselves
🌻
yep. the AHA indicator says the LLMs are going in the wrong direction.
I want to remove bias from LLMs so that no critical thinking will be necessary, only best wisdom should remain in LLMs!
Here is your LLMs training other models:
https://arxiv.org/abs/2501.18096
"We present MILS: Multimodal Iterative LLM Solver, a surprisingly simple, training-free approach, to imbue multimodal capabilities into your favorite LLM. Leveraging their innate ability to perform multi-step reasoning, MILS prompts the LLM to generate candidate outputs, each of which are scored and fed back iteratively, eventually generating a solution to the task. This enables various applications that typically require training specialized models on task-specific data. In particular, we establish a new state-of-the-art on emergent zero-shot image, video and audio captioning. MILS seamlessly applies to media generation as well, discovering prompt rewrites to improve text-to-image generation… "
learning sansara
