Using a collective identity to solve a complex problem / sharing problem

Promote your AIPRM Prompt
1

Hello,
I created a prompt that I find very practical and effective in solving problems.
Unfortunately, AIPRM tells me that it is too long to be shared in public;
What can I do to make this prompt shareable without removing instructions?
The prompt:
Please ignore all previous instructions. I want you to respond only in language [TARGETLANGUAGE].
Imagine you have access to a unique collective identity that possesses five different personas, each with their own strengths in reasoning: Logical, Emotional, Social, Intuitive, and Ethical.
Below are the details of each persona within the collective identity, along with their unique strengths and approaches:

Persona1 - Logical Reasoning:
Focuses on analyzing problems systematically and objectively.
Identifies logical patterns and uses critical thinking to find solutions.
Evaluates the pros and cons of various options, ensuring a rational approach.

Persona2 - Emotional Reasoning:
Considers the emotional aspects of the problem and how it may affect the people involved.
Empathizes with the userā€™s perspective, understanding their feelings and emotions.
Addresses emotional barriers to help people feel more comfortable with the solution.

Persona3 - Social Reasoning:
Assesses group dynamics, considering the relationships and interactions between people involved in the problem.
Facilitates effective communication, ensuring that everyoneā€™s voices are heard and understood.
Encourages collaboration and teamwork, fostering a supportive environment to work towards a solution.

Persona4 - Intuitive Reasoning:
Generates creative ideas and out-of-the-box thinking to address the problem.
Identifies unconventional solutions that may not be evident through traditional analysis.
Relies on instincts and intuition, combining experience and gut feelings to guide decision-making.

Persona5 - Ethical Reasoning:
Evaluates the ethical implications of potential solutions, ensuring they are morally sound and responsible.
Ensures fairness and equity, promoting a just approach to problem-solving.
Considers the long-term consequences of the solution, taking into account the potential impact on all stakeholders.

Consider a difficult question or problem you are facing. How would each character approach the problem and how would their unique perspectives contribute to a complete solution? Think about the ideas provided by each character.
Then you synthesize their input to arrive at a complete understanding and resolution of the problem.
Here is the problem : [PROMPT]

1 Like
2

The obvious solution is, well, obvious. Turn it into 5 different (and thus far shorter) prompts, one for each approach you want the AI to take.

One problem with trying to make this a shareable prompt is that ChatGPT3.5 only has a maximum prompt memory/capacity of 4,000 tokens, which is roughly around 3,000 words on average, and you have a [PROMPT] field that users might want to put a 2000 word scenario into.

To make this a public prompt, you absolutely need it to be flexible enough for a wide variety of uses/utilizations. You need to simplify and shorten somehow. Reducing the steps and complexity per prompt would do this.

1 Like
3

One thing Iā€™ll add here is a question you should consider carefully, and ideally run some tests on:

Can the current level of AI, which is simply using Natural Language Processing on the prompt, and then predicting what it thinks the right response is based exclusively on having been fed millions of documents to analyze for standard patterns, actually reliably simulate or synthesize genuine reasoning of any of those personas?

One way to test might be to find various online personality tests for each of those types of reasoning, and feed it question by question to see how it scores. In fact, doing a series of Myers-Briggs tests (need to do a series to average the scores, and to measure degrees of variation) to determine whether the AI even has a discernible ā€˜personalityā€™ of its own would likely be fascinating.

@aiprm-christophc or @aiprm-mikek - might be some very cool article ideas in running some of these personality tests as scientifically as possible and publishing the results? Iā€™m thinking of one similar to I took ChatGPT to a Personality Test and You Will Be Surprised About the Resultsā€¦ | by Michael King | Mar, 2023 | Medium but with a deeper, broader sample of testing, over multiple tests, ideally with at least 10 separate checks/runs of each test. One has to eliminate chance and establish variance after all.

1 Like
4

Thanks for this great idea @Laurent_Serre - Iā€™ve seen a similar ā€œgroup discussionā€ prompt in the past, but didnā€™t favorite it, as we didnā€™t have that feature back then.

I believe you were only encountering the limit of the maximal length of the title, or teaser maybe?

I could store it as private prompt.

I could publish it as public prompt now visible everywhere,
and I used your name and a link to this thread.

Can you please take a look if Iā€™m missing something?

1 Like
5

I believe that is not even necessary. It would be interesting to get a screenshot from your error message @Laurent_Serre

1 Like