From Poe to AI, Part Two

Welcome to Celebrate Creativity - previously celebrate poe - this is episode 401 - From Poe to AI, Part Two - In this podcast episode, I wanna be getting away from Poe for now and delving more into AI. And now when I talk about AI or AI bots, just think about websites that use AI can you can ask a question - more about that later.

Now, opening a play on Broadway certainly is a demanding undertaking.  And I want to start today's podcast by briefly mentioning two people who helped open plays on Broadway in their 90s. First, Adrienne Kennedy of Pittsburgh, Pennsylvania, born in 1931, is an American playwright known for her unique and surreal theatrical style. She grew up in Cleveland, Ohio, and later attended Ohio State University, where she earned a B.A. in education in 1952. Her plays frequently center Black women and delve into the impact of racism and societal pressures. In 2022, at the age of 91, Adrienne Kennedy made her Broadway debut as a playwright with Ohio State Murders. This play, written decades earlier, draws on Kennedy's own experiences as a student at Ohio State University in the late 1940s and early 1950s, highlighting the systemic racism she encountered.  

And John Harold Kander - born in 1927 - along with his partner lyricist Fred Ebb.) wrote the scores for 15 musicals, including Cabaret (1966) and Chicago (1975.), He and Ebb also wrote the standard "New York, New York" (officially known as "Theme from New York, New York"). When he was 94 he wrote the music for a Broadway musical version of New York New York - a play that received 11 Tony nominations and four wins.

Now I want to continue from last week and jump back to the subject of doing historical research with AI.  And for the rest of this episode, I am largely going over comments from Gemini.

Thank you for experiencing Celebrate Creativity.

Transcript

Welcome to Celebrate Creativity - previously celebrate poe - this is episode 401 - From Poe to AI, Part Two - In this podcast episode, I wanna be getting away from Poe for now and delving more into AI. And now when I talk about AI or AI bots, just think about websites that use AI can you can ask a question - more about that later.

Now, opening a play on Broadway certainly is a demanding undertaking.  And I want to start today's podcast by briefly mentioning two people who helped open plays on Broadway in their 90s. First, Adrienne Kennedy of Pittsburgh, Pennsylvania, born in 1931, is an American playwright known for her unique and surreal theatrical style. She grew up in Cleveland, Ohio, and later attended Ohio State University, where she earned a B.A. in education in 1952. Her plays frequently center Black women and delve into the impact of racism and societal pressures. In 2022, at the age of 91, Adrienne Kennedy made her Broadway debut as a playwright with Ohio State Murders. This play, written decades earlier, draws on Kennedy's own experiences as a student at Ohio State University in the late 1940s and early 1950s, highlighting the systemic racism she encountered.  

And John Harold Kander - born in 1927 - along with his partner lyricist Fred Ebb.) wrote the scores for 15 musicals, including Cabaret (1966) and Chicago (1975.), He and Ebb also wrote the standard "New York, New York" (officially known as "Theme from New York, New York"). When he was 94 he wrote the music for a Broadway musical version of New York New York - a play that received 11 Tony nominations and four wins.

Now I want to continue from last week and jump back to the subject of doing historical research with AI.  And for the rest of this episode, I am largely going over comments from Gemini.

Now While AI won't hand you primary sources, it can help identify types of primary sources relevant to a topic. "What kinds of primary documents would shed light on the conditions of factory workers in early industrial England?" (You'd still need to go to archives/digital libraries to access them).  

Specific Tips for Older Adults and Public Domain Research

Emphasize Accessibility: Highlight that these tools are often free to use at a basic level (e.g., ChatGPT's free tier, Google Gemini).  And personally, I currently use Gemini more than anything else.  I'm I'm afraid I would get sucked into paying for chat GPT if I really got into it, and I find Gemini.ai to be more direct.

Leveraging Existing Knowledge: Encourage listeners to use their lifetime of knowledge to ask more precise questions. Their wisdom can guide the AI to better results. For example, instead of "Tell me about Poe," a knowledgeable listener might ask, "Analyze the influence of German Romanticism on Poe's early poetry."

The "Curiosity Enhancer": Position AI as a tool that fuels curiosity, allowing users to dive down rabbit holes and explore tangential interests sparked by their main research.

Public Domain Advantage: Reinforce that since 19th-century works are in the public domain, you can feed excerpts or even full texts (within character limits) into an AI bot for analysis without copyright concerns. This is a huge benefit

Ethical Considerations (briefly): While verification is key, also touch upon responsible use. Don't present AI-generated content as your own original analysis without proper attribution or clear explanation that it's a tool you used. This also aligns with academic integrity.

I then asked Gemini for some more suggestions for my podcast.

Here are some more suggestions to enrich your "Celebrate Creativity" podcast, focusing on older creators, AI, and the rich public domain of the 19th century:

Broadening "Celebrate Creativity" - Beyond Poe & Whitman

While Poe and Whitman are your foundation, a broader scope allows for incredible diversity.

Episode Ideas:

"The Unsung Scribes: Rediscovering Lost Voices of the 19th Century":

Focus on a lesser-known but significant 19th-century author whose work is in the public domain. This could be a woman writer, an abolitionist writer, an early science fiction author, or someone from a marginalized community whose work offers a fresh perspective.

AI Integration: How can listeners use AI to discover these lesser-known authors? For example, "Ask AI for 3 female American poets from the 19th century who are not Emily Dickinson." Or "Generate a summary of early American utopian fiction."

Creative Prompt: Challenge listeners to read a work by one of these authors and share their "aha!" moments.

"The Art of Aging: Wisdom, Creativity, and the Long Arc of a Life":

Academic Insight: Briefly touch upon studies about neuroplasticity and the brain's ability to continue learning and creating throughout life  So I briefly delved into it.  And Gemini replied with:

For too long, there was a prevailing myth that once we reached adulthood, our brains were "hardwired" and largely incapable of significant change. Neuroplasticity shatters that idea. It's the brain's remarkable ability to reorganize itself by forming new neural connections and strengthening existing ones throughout life. Think of it as your brain being a dynamic, adaptable landscape rather than a fixed, rigid structure.

How it Works: "Neurons that Fire Together, Wire Together"

At its core, neuroplasticity involves several key mechanisms:

Synaptic Plasticity: This is perhaps the most famous aspect, often summarized by the phrase "neurons that fire together, wire together." Every time you learn something new, practice a skill, or even have a novel thought, the connections (synapses) between neurons involved in that activity are strengthened. The more often these connections are used, the stronger and more efficient they become, like a well-trodden path in a forest. Conversely, connections that aren't used can weaken or be "pruned" away.

Neurogenesis: While less widespread than synaptic changes, research has shown that new neurons can actually be generated in certain areas of the adult brain, particularly in the hippocampus, which is crucial for memory and learning.

Functional Reorganization: If one area of the brain is damaged (e.g., from a stroke), other areas can sometimes take over its functions, demonstrating the brain's incredible capacity to compensate and adapt.

Lifelong Learning: The Ultimate Brain Workout

The implications of neuroplasticity for lifelong learning are profound:

It's Never Too Late to Learn: Whether it's picking up a new language at 70, mastering a musical instrument in your 50s, or diving into a completely new field of study, your brain is fundamentally equipped to acquire new knowledge and skills at any age.

Continuous Improvement: Every new experience, every challenging puzzle, every new piece of information you absorb actively reshapes your brain, making it more efficient and capable. This isn't just about retaining facts; it's about building more robust neural networks for processing information, problem-solving, and critical thinking.

 Neuroplasticity and Creativity: The Unseen Connection

This is where it gets particularly exciting for your podcast audience:

Connecting Disparate Ideas: Creativity often involves making novel connections between seemingly unrelated concepts. Neuroplasticity is the biological mechanism that allows your brain to form and strengthen these "unconventional" neural pathways. When you engage in creative thinking, you're literally building new bridges between different regions of your brain.

Adapting and Evolving Creative Processes: A creative person isn't stuck with one way of thinking or creating. Neuroplasticity enables artists, writers, musicians, and innovators to continuously adapt their approaches, learn from failures, incorporate new influences, and evolve their craft. The brain can "rebound" from creative setbacks by gleaning lessons and adjusting.

Enhanced Problem-Solving: Creativity is inherently linked to problem-solving. By constantly challenging your brain with new ideas and artistic endeavors, you're enhancing its ability to explore multiple solutions and think outside the box.

Dopamine and Motivation: Engaging in creative activities can also trigger the release of neurotransmitters like dopamine, which not only amplify motivation but also play a role in synaptic strengthening, further reinforcing the learning and creative process.

Practical Takeaways for Your Listeners:

Embrace Novelty: Actively seek out new experiences, learn new skills, read diverse materials, and engage in different forms of art. Novelty is a powerful stimulant for neuroplasticity.

Challenge Yourself: Don't shy away from tasks that feel difficult or outside your comfort zone. These are the moments when your brain is working hardest to adapt and grow.

Practice Deliberately: Whether it's a creative pursuit or a new skill, consistent and focused practice strengthens neural connections. "Practice makes permanent," as the saying goes.

Cultivate a Growth Mindset: Understanding neuroplasticity empowers individuals to believe in their own capacity for growth and change. This belief itself can encourage more engagement and effort, leading to greater neural rewiring.

Prioritize Brain Health: Good sleep, nutrition, exercise, and managing stress all contribute to a healthy brain environment that supports neuroplasticity.

In essence, neuroplasticity is the biological assurance that our capacity for learning and creativity isn't a fixed well that eventually runs dry. It's a perpetually regenerating spring, constantly refreshed by our experiences, efforts, and curiosity. This is incredibly empowering for anyone looking to live a more fulfilling, intellectually vibrant, and creatively rich life, no matter their age. I think those are some incredibly optimistic words - thoughts that lead to a great outlook on life.  And I really intend to deal into neuroplasticity a great deal in future episodes.

But back to some of Geminis other suggestions for this podcast:

Listener Stories: Encourage listeners to share their own creative pursuits or how they've re-engaged with a passion in later life.

"Victorian Visionaries: Science, Spirituality, and the Imagination":

Explore how science and new discoveries (e.g., Darwin, early photography, electricity) influenced 19th-century writers and their creative output.

AI Integration: Ask AI to explain the basic scientific understanding of the universe in 1850. Or "How did the invention of the telegraph influence literature?"

"From Page to Pixel: Adapting Public Domain Works for New Media":

Creative Prompt: Challenge listeners to imagine how they would adapt a specific public domain work (e.g., The Tell-Tale Heart, a Whitman poem) for a modern audience or a new medium. How would AI assist them in this? ("Generate 5 modern settings for Poe's 'The Fall of the House of Usher.'")

Pick a specific 19th-century poem (e.g., one of Whitman's shorter works, a Longfellow ballad, a Dickinson poem).  Use AI to identify poetic devices (alliteration, meter, rhyme scheme), as well as analyze word choice and its effect.

Compare different critical interpretations.

Even suggest alternative wordings (just for fun, to see how the AI "thinks" about poetry, not to claim it's better). This shows AI as a tool for deeper analysis, not just quick answers.

Dedicate an episode or two on how to ask better questions ("prompts") to AI bots.  And I already done one that I haven't broadcast about Edgar Allan Poe applying for a job with a pet rental company that specializes in black cats - sort of an example of the use of role-playing in a prompt. 

More seriously, be clear and specific: "Don't just say 'Tell me about Poe.' Say 'Analyze the use of symbolism in Edgar Allan Poe's 'The Raven' focusing on the raven itself and the bust of Pallas.'"

Role-Playing: tell AI to "Act as if it were a 19th-century literary critic and review Walt Whitman's Leaves of Grass."

Iterative Prompting: Show how you refine a question based on the AI's initial response. "That's good, now expand on point #3, providing specific examples from the text."

Asking for Sources: While AI sometimes struggles with direct citations, you can prompt, "List potential primary sources that would support this argument."

This empowers listeners to get more out of their AI interactions.  Again this might seem like a bit dense now, but the purpose is to make them much easier to understand,

"The Future of History: How AI is Changing Our Understanding of the Past":

Discuss how large language models (LLMs) are enabling new forms of historical analysis, such as identifying patterns in vast amounts of text that a human eye might miss.

Now, for the rest of this episode, I want to delve into exactly what a large language model is - really important to understanding this stuff.  

So we have large language model.

Large: This refers to the immense scale of these models. They are trained on truly colossal datasets of text (and often other forms of data such as images and audio, making them "multimodal" in some cases). We're talking billions or even trillions of words, drawn from books, articles, websites, conversations, and more. This "largeness" also extends to the number of internal parameters (such as adjustable dials) the model uses to understand and generate language – often in the hundreds of billions.

Language: As the name suggests, LLMs are primarily designed to understand, process, and generate human language. They learn the patterns, grammar, semantics, and even nuances of language through their extensive training.

Model: In the context of AI, a "model" is a mathematical construct that has learned to identify patterns and make predictions based on the data it was trained on. LLMs use a specific type of artificial neural network called a transformer architecture, which is particularly effective at handling text.

And what follows is a simplified version of how large language models work.

Imagine feeding a vast library of all human written knowledge into a super-advanced computer program. The program doesn't "understand" in a human sense, but it learns statistical relationships between words and phrases.

At a high level, an LLM's training process involves:

Pre-training: The model is fed massive amounts of text and learns to predict the next word in a sentence, or fill in missing words. By doing this over and over, it develops an incredibly sophisticated understanding of how language is structured and how words relate to each other.

Fine-tuning: After the initial pre-training, LLMs are often further refined (fine-tuned) for specific tasks, like answering questions, engaging in conversations, or following instructions. This often involves techniques such as "reinforcement learning from human feedback" to make them more helpful, honest, and harmless.  Now Because of their extensive training, LLMs can perform a wide range of language-related tasks, including:

Generating text: Writing articles, stories, poems, emails, code, and more.

Answering questions: Providing informative responses to a vast array of queries.

Summarizing text: Condensing long documents into shorter, coherent summaries.

Translating languages: Converting text from one language to another.

Chatbots and virtual assistants: Powering conversational AI systems.

Content creation: Helping with brainstorming and generating creative ideas. For example, My other half just had Gemini rewrite his résumé to sound better and have more active verbs.

Code generation and debugging: Assisting programmers in writing and fixing code.

In essence, an LLM is a powerful AI system that has learned the intricate patterns of human language to such an extent that it can understand, generate, and interact with text in remarkably human-like ways. They are the technology behind many of the "generative AI" applications we see today, like ChatGPT and Gemini.

Oh yes, there is one thing more I want to reiterate - the possibility that an AI will “hallucinate” or tell use something that is not accurate using AI does not eliminate the need for human verification and critical thinking. AI is a tool, not a substitute for scholarly rigor.

Join Celebrate Creativity for episode 402 - The Future of History

Sources include ChatGPT and Gemini.ai

Thank you for listening to celebrate Poe - I mean Celebrate Creativity.