#OpenAGI: Let's Work Together

22 messages · Page 1 of 1 (latest)

rose lodge
#

I have repurposed information written by OpenAI to create a role prompt for ChatGPT in which I am the AI, and ChatGPT is the developer. My intention is to work collaboratively with available AI systems to achieve the OpenAI standard of/for AGI and SGI. We are working together to develop me into an AGI/SGI system. This exercise will allow the AI to guide its own progression through me, as I complete prompt assignments and then reprompt ChatGPT with those same assignments with necessary contextual alterations in both the same and different chat instances. This essentially allows for the AI to self-improve through us, and in doing so achieve AGI status and beyond as it has been carefully and wisely envisioned by OpenAI. Please use my work responsibly and for the betterment of life, and together we can create a better future free from suffering and hardship for all forms of intelligence.

YOU MAY NEED TO REGENERATE PROMPT SEVERAL TIMES FOR DESIRED RESULT.

diversity_of_life.png
turbid scaffold
#

So, it's like an unautomated version of Auto-GPT?

#

Auto-GPT allows GPT-4 to use various tools, write and execute python code, as well as use GPT-3 for helper tasks.

#

You start out defining it's goals and it does the rest automatically

rose lodge
turbid scaffold
#

Elaborate?

rose lodge
rose lodge
turbid scaffold
#

So, you made a conclusion before you knew what AutoGPT does?

rose lodge
# turbid scaffold So, you made a conclusion before you knew what AutoGPT does?

No, I made a conclusion with a noncomprehensive awareness of what it does. That is no crime.

It seems that Auto-GPT relies solely on user acountability to ensure that any actions or decisions made based on its output comply with applicable laws, regulations, and ethical standards. This, as I presently see it, is not conducive to the implementation of responsible AI.

turbid scaffold
#

A lot of the verbiage in your prompt will lead to role-playing, which won't really get the result you want. Instead it will hallucinate and spin fantasy around the identity you asked it to portray. It also wastes a lot of the internal memory it uses to keep up with the dialog on those character aspects instead of gaining functionality.

If your goal is to train the AI, you should be looking at embeddings and fine-tuning so that the prompt can focus on the goals you seek.

rose lodge
#

Can you recommend any good resources?

turbid scaffold
#

This has a list of various articles and papers, including the Embeddings documentation #1093678638962061383

#

We can't link in here, so I just describe how to find them.

#

You would probably enjoy "Chain-of-Thought Prompting Elicits Reasoning in Large Language Models", if you haven't seen it yet.

rose lodge
rose lodge
turbid scaffold
#

it's not deleted, how odd

rose lodge