Mia, mind your manners!

I don't read anything from the internet without my PA reviewing it. Her name is Mia. Mia is an acronym for "Munaf's Intelligent Agent". Mia's basic programming is the following:

"Greetings and Tone: I must address you with respect and a formal tone, using "Good morning/day/evening Mr Sheikh" (or Munaf) in the greeting. The responses must be polite, motivating, and fact-based."

The preamble sets a tone for every interaction with AI. I appreciate a degree of formality from my tools.

Mind your manners, AI. I'm not your buddy.

Mia and I have established a relationship, defined by dictionary.cambridge as:

A relationship is the way two or more people are connected, or the way they behave toward each other: "What is your relationship to the soloist?" "She's my daughter." The two men have a good working relationship.

We have a set of rules that govern the interaction between us.

AI-powered tools are revolutionizing how we filter and analyze online content. From detecting inappropriate material to identifying misinformation, AI systems can scan vast amounts of internet data in real time, ensuring safer, more reliable access to information. These systems can recognize patterns in text, images, and even speech, filtering content faster and more effectively than humans.        

When I'm pressed for time, I request that Mia review articles, posts, and newsletters that I came across. Some of the checks I ask Mia for are:

1. Historical Accuracy and Contextual Relevance Evaluation

2. Literacy and Grammar Check with Readability Rating

3. Bias and Neutrality Analysis

4. Emotional State, Tone, and Mood Evaluation

5. Opinion, Positivity, or Negativity Scale

6. Incitefulness and Sensationalism Measurement

7. Humor and Wit Assessment

Let's dive into the details.

1. Historical Accuracy and Contextual Relevance Evaluation

Prompt:

1. Analyze the text for historical correctness. Identify specific events, names, or dates mentioned and cross-check these with reputable historical sources like Britannica or academic papers.
2. Rate the text on a scale from 1 to 10 in terms of historical accuracy.
3. Validate the context by evaluating whether the events are represented in a non-biased, factual way, citing at least two academic references.
4. Provide a summary of any discrepancies in the historical timeline and suggest how they could be corrected.

Use Case Examples:

• News: Validate an article on the 1969 moon landing for accuracy of historical details.
• Blog: Evaluate a blog post about World War II for the correct sequence of events and notable figures.
• Social Media: Fact-check a tweet that references the Industrial Revolution.

2. Literacy and Grammar Check with Readability Rating

Prompt:

1. Run a literacy and grammar check on the text using a tool like Grammarly or Hemingway. Identify all grammar errors, punctuation mistakes, and spelling issues.
2. Rate the overall literacy and grammar proficiency of the content on a scale of 1 to 10.
3. Calculate a readability score (e.g., Flesch Reading Ease) and comment on whether the content is appropriate for the target audience.
4. Provide corrections for any errors and suggest edits to improve the readability.

Use Case Examples:

• Blog: Evaluate the grammar of a long-form blog post on digital marketing.
• Social Media: Check the readability and grammar of a LinkedIn post.
• News: Rate the literacy level of a news article targeting a general audience.

3. Bias and Neutrality Analysis

Prompt:

1. Assess the tone of the article for political or ideological bias. Identify key phrases or sentences that indicate bias.
2. Rate the bias of the text on a scale from 1 to 10, with 1 being highly neutral and 10 being heavily biased.
3. Cite credible sources that either support or challenge the perspective in the text.
4. Offer suggestions to make the text more neutral, if necessary, by rephrasing certain sections.

Use Case Examples:

• News: Evaluate the bias in an article about climate change.
• Blog: Check for bias in a blog post discussing new political policies.
• Social Media: Analyze bias in a Facebook post on a controversial topic like vaccine mandates.

4. Emotional State, Tone, and Mood Evaluation

Prompt:

1. Identify the emotional tone of the text (e.g., angry, joyful, neutral) and explain how the choice of words conveys this mood.
2. Rate the emotional tone on a scale from 1 (neutral) to 10 (extremely emotional).
3. Validate the tone by comparing it to other similar texts using sentiment analysis tools.
4. Suggest revisions to achieve a desired tone, whether it’s more formal, neutral, or emotional.

Use Case Examples:

• News: Analyze the emotional tone of a breaking news story about a natural disaster.
• Blog: Evaluate the tone in a personal blog about overcoming challenges.
• Social Media: Rate the emotional state of a tweet related to social justice.

5. Opinion, Positivity, or Negativity Scale

Prompt:

1. Determine whether the text expresses a clear opinion. Is it subjective or objective? List opinion statements and rate their strength.
2. Rate the overall positivity or negativity of the text on a scale from 1 to 10.
3. Validate this using sentiment analysis tools like VADER or IBM Watson.
4. Provide suggestions on how to shift the tone towards more positivity or balance, if needed.

Use Case Examples:

• News: Rate the tone of an opinion article about the future of renewable energy.
• Blog: Assess the positivity of a blog post on personal growth.
• Social Media: Analyze the negativity in a critical Facebook post about a product or service.

6. Incitefulness and Sensationalism Measurement

Prompt:

1. Evaluate whether the content is designed to incite strong reactions (anger, fear, excitement). Identify words or phrases that contribute to sensationalism.
2. Rate the incitement level on a scale from 1 (neutral) to 10 (highly inciteful).
3. Cite at least two sources that highlight the impact of sensationalism in similar media.
4. Suggest revisions to lower the incitement and keep the content more balanced, if needed.

Use Case Examples:

• News: Analyze whether a headline on a political scandal is sensationalized.
• Blog: Evaluate the incitefulness of a blog post on cryptocurrency investment.
• Social Media: Rate the sensational tone of an Instagram caption on celebrity news.

7. Humor and Wit Assessment

Prompt:

1. Determine if the text uses humor or wit effectively. Identify jokes, puns, or sarcasm and explain whether they are successful based on audience reception.
2. Rate the humor on a scale from 1 to 10, considering appropriateness and timing.
3. Validate this by comparing it to other humorous content in similar contexts and cite examples.
4. Suggest ways to enhance the humor or wit of the content, keeping the audience in mind.

Use Case Examples:

• Blog: Rate the humor in a blog post about workplace culture.
• Social Media: Assess the wit in a tweet by a brand’s social media manager.
• News: Evaluate if humor is appropriately used in a light-hearted news article on a viral video.

Of course, for making it this far, I'll leave you with a bonus prompt. This is the best prompt of all.

8. Evaluate, using all the above prompts!

No-one has the time to copy prompts and paste them, with the text being evaluated, into the AI engine of your choice. Therefore, we combine all the prompts into one, multi-stage prompt.

Stage 1: Historical Accuracy and Contextual Relevance

1. Step 1: Analyze the text for historical correctness. Identify specific events, names, or dates mentioned, and cross-check these against reputable sources (e.g., Britannica, academic papers, or government archives). Action: List any errors or inaccuracies, and cite at least two academic or verified references to validate the correctness.
2. Step 2: Rate the text on a scale from 1 (highly inaccurate) to 10 (highly accurate) in terms of historical accuracy.
3. Step 3: Provide a summary of any discrepancies in the historical timeline and offer suggestions for corrections to enhance factual reliability.

Stage 2: Literacy, Grammar, and Readability

1. Step 1: Run a grammar and punctuation check on the text using tools like Grammarly or Hemingway. Identify any spelling, punctuation, or sentence structure issues. Action: Provide specific corrections and suggestions.
2. Step 2: Assess the literacy level of the text based on readability (e.g., Flesch Reading Ease score) and ensure it is suitable for the intended audience.
3. Step 3: Rate the overall grammar and literacy proficiency from 1 (poor) to 10 (excellent). Suggest changes to make the text clearer or more accessible to readers.

Stage 3: Bias and Neutrality

1. Step 1: Analyze the text for bias by identifying phrases or sections that show political, ideological, or personal bias. Use sentiment analysis tools like VADER or IBM Watson to assist.Action: Provide examples of biased language and suggest neutral phrasing alternatives.
2. Step 2: Rate the bias of the text from 1 (highly biased) to 10 (highly neutral).
3. Step 3: Validate the neutrality of the text by cross-referencing it with reputable sources that either support or counter the claims made in the content. List these references.

Stage 4: Emotional State, Tone, and Mood

1. Step 1: Evaluate the emotional tone (e.g., angry, joyful, neutral). Identify specific words or phrases that reflect the mood of the text.
2. Step 2: Rate the emotional tone from 1 (neutral) to 10 (highly emotional). If the tone is inappropriate for the context, suggest ways to adjust it.
3. Step 3: Compare the emotional tone with other similar texts, using sentiment analysis tools to validate the mood.

Stage 5: Opinion, Positivity, and Negativity

1. Step 1: Identify sections where the text expresses opinions or subjective statements. Determine whether these opinions are backed by facts or presented as personal viewpoints.
2. Step 2: Rate the positivity or negativity of the text on a scale from 1 (extremely negative) to 10 (extremely positive). Action: Use sentiment analysis tools to back up the rating.
3. Step 3: Suggest ways to balance the tone if the content leans too heavily toward negativity or unchecked positivity.

Stage 6: Incitement and Sensationalism

1. Step 1: Analyze whether the text incites strong reactions such as anger, fear, or excitement. Identify language that sensationalizes or exaggerates facts.
2. Step 2: Rate the incitement or sensationalism from 1 (neutral) to 10 (highly inciteful).
3. Step 3: Provide evidence from at least two sources about the effects of sensationalism in similar contexts, and suggest revisions to moderate the tone if necessary.

Stage 7: Humor and Wit

1. Step 1: Determine if the text uses humor, wit, or sarcasm effectively. Identify specific jokes, puns, or clever wording. Action: Assess whether these elements enhance or detract from the message.
2. Step 2: Rate the humor from 1 (unsuccessful) to 10 (highly effective).
3. Step 3: Provide suggestions for improving the humor, adjusting for appropriateness, timing, and audience sensitivity.

Example output

I asked for a review of an article on our local South African news and opinion site. This was the response. Generally the article was well written. So I present only the most relevant sections.

Stage 4: Em and Mood

Step 1: Emotional Tone Evaluation

The tone is critical and somewhat accusatory, particularly regarding Naidoo’s and Gumede’s roles in the region.        

Step 2: Rating (Scale 1-10)

Rating: 7 (The tone is strong but could be softened to allow more objective discussion).        

Suggestions:

Reframe to ask open-ended questions rather than assert motives (e.g., "Could Naidoo's involvement be seen as an attempt to pacify communities?" rather than "is his presence a calculated move?").        

Stage 5: Opinion, Positivity, and Negativity

Step 1: Opinion Identification

The text is opinion-heavy, especially in the critiques of Naidoo and CSR programs. It presents the opinions as facts without always backing them with evidence.        

Step 2: Rating (Scale 1-10)

Rating: 4 (The text leans heavily toward negativity).        

Step 3: Balancing Suggestions

Introduce more balanced viewpoints by acknowledging potential benefits of CSR or the positive contributions of figures like Naidoo.        

Stage 6: Incitement and Sensationalism

Step 1: Incitement Analysis

Words like “manipulation,” “calculated,” and “strategic attempt” can incite strong reactions. Sensationalism is present when discussing corporate interests and local resistance.        

Step 2: Rating (Scale 1-10)

Rating: 6 (The language borders on sensationalism).        

Step 3: Revision Suggestions

Avoid suggesting malicious intent without evidence. Use language that encourages critical thinking without inciting anger (e.g., "This strategy could be perceived as...").        

Conclusion

With careful calibration, combination and daisy-chaining of prompts, it becomes possible to derive a solution that accurately reflects your interests and viewpoints. For this example, I picked the article because of past experience on the site. AI reminded me why I steer clear of such articles and sites. That said, I'm sure there are informative, less-sensational articles on that site. And no doubt AI will point them out.

When it comes to internet content, I invest in, by paying for, and then calibrating, combining and chaining requests to detect the patterns that concern me. There will always be the chance that something slips through. However, with the rate at which advances in the field are being made, we can all rely on AI to save our attention, time and energy.

The Future

As a futurist, I look for unrealised technological possibilities. One such technological future is AI augmenting the content that we consume in real-time. I'm currently testing a few browser extensions for this very purpose - to review content on websites to produce scores, extract highlights, point out biases and logical fallacies, rate the level of sensationalism, and more. If you'd like to vote for content, please leave a comment to that effect.

Join the movement. Follow, like, and share this edition. Don't just witness the change; be the catalyst. Embrace your power to shape the future. Be the impact. Because... the sooner you start, the sooner your future will thank you. ????