Generative AI Tools in 2024 (3 of n) Special Purpose AI Tools

First published at ScenarioPlans.com (DelphiPlan.com). Reprinted here with permission.

Conversation about GenAIs and Special Application AIs

This is 3 of n articles on the top GenAI tools in 2024.

Pushing the Boundaries: Can Your Free GenAI Rise to the Challenge?

We all love our go-to GenAI tools, but have you ever wondered just how far you can push their capabilities? In this upcoming blog post, we’ll take a popular free GenAI and try something it might not be designed for – think building complex tables, drawing intricate flowcharts, crafting web pages or programs, or generating stunning graphic images. Can it pull it off, even without those fancy paid features? Get ready for some unexpected results when you try this, and see if your free GenAI can surprise you!

Graphic produced by Mermaid AI based on results from Gemini Advanced; Background Image by DALL-E (2024, Aug.). Prompts and modifications by E. Hall.

Rationale: I have been pushing the limits of my 5 big general purpose GenAIs to produce the final table for a SWOT matrix. I finally give explicit instructions to give the external factors on the top, starting with column 2 for the Opportunities and column 3 for the Threats (or Risks/Roadblocks for Nonprofits). Then after the OT factors have been listed, list the Strengths and Weaknesses in column 1 with the initiatives/strategies inside the matrix. I usually just cut and paste into a table (in Excel). But, GenAIs will produce the instructions to build a table which can then be plugged into other applications, or special purpose GenAIs. Very cool.

Graphics, flow charts and infographs have been very problematic. ChatGPT passes you off to DALL-E with instructions and the final graphic will tend to have smudges, misspellings and inconspicuous mistakes. Sometimes the instructions can be in the form of code (or metacode). You can cut and paste the code into a specialized program.

A recent session produced some flow-diagram code because the GenAI couldn’t actually produce the final product of what it produced.

“COOL. Thanks a lot. What am I gonna do with this”, I ask.

Answer: There several specialized programs that can use the meta code including … Mermaid AI …

So, 5 minutes later, MermaidChart.com is up in my browser window in some kind of free mode. Paste in the code. And wa-la. Not the graphic chart I was looking for, but a cool chart, just the same. You can edit the code and it will display in the display window, or you can edit the display window and it will change the code.

The chart itself is rather interesting. It shows MS Copilot riding at the top of various AIs including, of course, ChatGPT which is its underlying preprocessor. You get access to the 4o paid version of ChatGPT for free when you use Copilot, and you get Copilot for free when you have MS Office kinda stuff.

But the POWER of copilot is working it into the office flow and workflow. It can do really smart things with any types of data (internal and/or external), emails, planning, etc.

Chats with MS Copilot (2024, Aug)

Sent by you:

What are several of the most important or most productive ways to use MS Copilot? Bullet points please.

Sent by Copilot:

Certainly! Here are several productive ways to use Microsoft Copilot:

• Content Creation: Generate outlines, chapters, and graphics for workbooks, reports, and presentations.
• Strategic Planning: Assist with SWOT analyses, trend identification, and other strategic planning tasks.
• Data Analysis: Analyze data sets, create visualizations, and generate insights.
• Task Automation: Automate repetitive tasks, such as data entry, scheduling, and email management.
• Research Assistance: Gather information, summarize articles, and provide references for various topics.
• Language Support: Translate text, proofread documents, and improve writing quality.
• Learning and Development: Provide explanations, tutorials, and practice exercises for various subjects.
• Meeting Support: Summarize meeting notes, generate action items, and schedule follow-ups.

Is there a specific area you’d like to explore further?

Sent by you:

What about in the office setting where Copilot can given internal data and such?

Sent by Copilot:

In an office setting, Microsoft Copilot can be incredibly useful when given access to internal data. Here are several productive ways to leverage Copilot:

• Document Generation: Create reports, proposals, and presentations using internal data.
• Data Analysis: Analyze sales figures, performance metrics, and other internal data to generate insights and visualizations.
• Meeting Summaries: Summarize meeting notes, generate action items, and track follow-ups based on internal meeting data.
• Email Management: Draft, organize, and prioritize emails using internal communication data.
• Project Management: Assist with project planning, task assignments, and progress tracking using internal project data.
• Customer Insights: Analyze customer feedback, support tickets, and sales data to identify trends and areas for improvement.
• HR Support: Generate employee performance reviews, training materials, and onboarding documents using internal HR data.
• Financial Reporting: Create financial statements, budget reports, and forecasts using internal financial data.

Would you like to dive deeper into any of these areas?

Sent by you:

What about learning, training and development where existing internal documents (SOPs) can be distilled with best practices to produce what’s needed: when needed, as needed?

Sent by Copilot:

Absolutely! Microsoft Copilot can be a powerful tool for learning, training, and development by leveraging existing internal documents (SOPs) and distilling them into best practices. Here are several productive ways to use Copilot in this context:

• Customized Training Materials: Generate training manuals, guides, and e-learning modules based on internal SOPs and best practices.
• On-Demand Learning: Provide employees with quick access to relevant training materials and best practices as needed.
• Interactive Tutorials: Create step-by-step tutorials and interactive learning experiences using internal documents.
• Knowledge Base Creation: Develop a centralized knowledge base with distilled best practices and SOPs for easy reference.
• Performance Support: Offer real-time support and guidance to employees during their tasks by referencing internal SOPs.
• Compliance Training: Ensure employees are up-to-date with compliance requirements by generating training materials from internal documents.
• Skill Assessments: Create quizzes and assessments to evaluate employee understanding of SOPs and best practices.
• Continuous Improvement: Regularly update training materials and best practices based on feedback and new internal data.

Would you like to explore any of these areas in more detail?

CODE (MetaCODE)

graph LR

subgraph General-Purpose Language Models

    ChatGPT[“ChatGPT <br> Versatile & conversational”]

    GPT4[“GPT-4 <br> Advanced & complex tasks”]

    Claude[“Claude <br> Safe & ethical”]

    Gemini[“Gemini <br> Real-time data”]

    Perplexity[“Perplexity <br> Accurate & concise answers”]

end

subgraph Specialized Tools

    Lumen5[“Lumen5 <br> Video creation”]

    Soundraw[“Soundraw <br> Music generation”]

    KeywordInsights[“KeywordInsights.ai <br> SEO & marketing”]

    ChatFlash[“ChatFlash <br> Customer engagement”]

end

subgraph Microsoft Ecosystem

    Copilot[“MS Copilot <br> Productivity assistant”]

end

%% Styling and positioning adjustments for better visual appeal

linkStyle default stroke-width:2px,fill:none,stroke:grey;

classDef mostUsed fill:#d0f0c0,stroke:#0f9d58;

classDef specialized fill:#cfe2f3,stroke:#201aa2;

classDef microsoft fill:#f4cccc,stroke:#b45f06;

class ChatGPT,GPT4,Claude,Gemini,Perplexity mostUsed;

class Lumen5,Soundraw,KeywordInsights,ChatFlash specialized;

class Copilot microsoft;

Q: Sometimes, often maybe, the General purpose GenAI will produce the instructions or code that can then be used with another AI program. Describe how that works and give a short list of AIs that can be used in each of the cases listed.

Gemini Advanced (2024, Aug)

Unlocking Multi-AI Synergy: When GenAI Writes the Code

In many scenarios, especially when tackling tasks beyond their core design, general-purpose GenAI tools act as the bridge to more specialized AIs. They excel at generating instructions, code snippets, or structured data, which can then be fed into other AI programs fine-tuned for specific tasks.

Let’s see how this plays out in some of the cases we mentioned:

1. Generating Tables:

• GenAI: Crafts detailed descriptions of the table structure, including column names, data types, and potential content.
• Specialized AI:
  • Table-building libraries (e.g., Pandas in Python): Interpret the instructions to create the table directly within the GenAI’s interface.
  • Data visualization tools (e.g., Tableau, Google Sheets): Take the structured data and build interactive, customizable tables.

2. Drawing Flowcharts:

• GenAI: Produces step-by-step process descriptions or even pseudocode outlining the flowchart’s logic.
• Specialized AI:
  • Flowchart generation tools (e.g., Lucidchart, Draw.io): Translate the textual description into visual flowcharts.
  • Diagrammatic reasoning engines: Understand the underlying logic and create more sophisticated, intelligent flowcharts.

3. Creating Web Pages:

• GenAI: Generates HTML, CSS, and even basic JavaScript code based on high-level descriptions of the webpage’s layout and functionality.
• Specialized AI:
  • Web development frameworks (e.g., React, Angular): Take the code and render it into interactive web pages.
  • Design AI tools: Help refine the aesthetics and user experience based on the generated structure.

4. Writing Programs:

• GenAI: Produces code snippets or entire programs in various languages based on natural language problem descriptions.
• Specialized AI:
  • Code completion and debugging tools: Help refine and complete the generated code.
  • AI-powered IDEs: Provide suggestions and optimizations based on the code’s purpose.

5. Creating Graphic Images:

• GenAI: Generates detailed text prompts describing the desired image.
  • Transform the prompts into visually stunning images.
  • Image editing tools: Allow further refinement and customization of the generated images.
• Specialized AI:
  •  Image generation models (e.g., DALL-E 2, Midjourney):

Remember: The effectiveness of this multi-AI approach depends heavily on the clarity of instructions the GenAI provides and the capabilities of the specialized AI receiving them. It’s a rapidly evolving field with exciting potential, so stay tuned for even more impressive collaborations between different AI tools!

#GenAI #IntellZine #RegenerativeAI #RDAI #ScenarioPlans #RapidDynamicAI

#NonprofitPlan #SBPlan #IPPlan #IntellZine

1. GenAI Prompt Guide – Stanford University –  uit.stanford.edu