Writing Advice - From CS Lewis to Generative AI - DeepSeek

 Saw a post on Facebook with the following content:

It was C S Lewis giving advice to a schoolgirl on writing. The post is:

TO A SCHOOLGIRL IN AMERICA, who had written (at her teacher’s suggestion) to request advice on writing.   14 December, 1959

It is very hard to give any general advice about writing. Here’s my attempt.

1. Turn off the Radio.
2. Read all the good books you can, and avoid nearly all magazines.
3. Always write (and read) with the ear, not the eye. You should hear every sentence you write as if it was being read aloud or spoken. If it does not sound nice, try again.
4. Write about what really interests you, whether it is real things or imaginary things, and nothing else. (Notice this means that if you are interested only in writing you will never be a writer, because you will have nothing to write about. . . .)
5. Take great pains to be clear. Remember that though you start by knowing what you mean, the reader doesn’t, and a single ill-chosen word may lead him to a total misunderstanding. In a story, it is terribly easy just to forget that you have not told the reader something that he wants to know—the whole picture is so clear in your own mind that you forget that it isn’t the same in his.
6. When you give up a bit of work don’t (unless it is hopelessly bad) throw it away. Put it in a drawer. It may come in useful later. Much of my best work, or what I think my best, is the re-writing of things begun and abandoned years earlier.
7. Don’t use a typewriter. The noise will destroy your sense of rhythm, which still needs years of training.
8. Be sure you know the meaning (or meanings) of every word you use.

I feed to DeepSeek for its comments:

DeepSeek: 

C.S. Lewis's letter to the schoolgirl offers timeless advice on writing, blending practical tips with philosophical insights. Here are some further reflections on his points:

1. "Turn off the Radio": Lewis emphasizes the importance of focus and minimizing distractions. In today’s context, this could extend to turning off social media, notifications, or any other digital interruptions. Writing requires deep concentration, and external noise can disrupt the flow of thought.

2. "Read all the good books you can, and avoid nearly all magazines": Lewis advocates for immersing oneself in quality literature. Good books expose writers to rich language, compelling narratives, and diverse perspectives, while magazines (especially those of his time, which were often superficial) might not offer the same depth. Today, this could also apply to discerning between high-quality content and clickbait or shallow online articles.

3. "Always write (and read) with the ear, not the eye": This is perhaps one of the most profound pieces of advice. Writing is not just a visual exercise; it’s an auditory one. The rhythm, flow, and musicality of language are crucial. If a sentence doesn’t sound right when spoken aloud, it likely won’t resonate with readers. This advice underscores the importance of reading drafts out loud during the editing process.

4. "Write about what really interests you": Authenticity is key. Writing fueled by genuine passion and curiosity will always engage readers more than writing that feels forced or insincere. Lewis also warns against the trap of being interested only in the act of writing itself—writers need substance, experiences, and ideas to draw from.

5. "Take great pains to be clear": Clarity is essential in communication. Writers often fall into the trap of assuming readers understand their perspective or the world they’ve created. Lewis reminds us to step back and consider the reader’s experience, ensuring that every detail necessary for understanding is included.

6. "When you give up a bit of work, don’t throw it away": This is excellent advice for dealing with creative frustration. Not every piece of writing will work immediately, but ideas often need time to mature. Revisiting old work with fresh eyes can lead to breakthroughs. This also speaks to the value of patience and persistence in the creative process.

7. "Don’t use a typewriter": While this advice might seem outdated in the age of computers, the underlying principle remains relevant. Writing tools can influence the creative process. For Lewis, the typewriter’s noise disrupted his rhythm. Today, writers might find that certain tools or environments help or hinder their flow. Experimentation is key.

8. "Be sure you know the meaning of every word you use": Precision in language is vital. Misusing a word can confuse readers or distort the intended message. This advice encourages writers to be deliberate and thoughtful in their word choices, which is especially important in an era where language is often used carelessly.

Overall, Lewis’s advice is a blend of practical discipline and creative freedom. He encourages writers to be intentional, patient, and authentic, while also reminding them to enjoy the process and stay true to their interests. His insights remain remarkably relevant, offering guidance not just for writing, but for any creative endeavor.

Here comes the more interesting past. I asked DeepSeek what other advice it could give.

 DeepSeek:

Certainly! Building on C.S. Lewis's advice, here are some additional thoughts and tips for aspiring writers:

1. Write Regularly, Even When You Don’t Feel Inspired

Writing is a skill that improves with practice. Set aside time to write consistently, even if it’s just a few sentences a day. Inspiration often comes during the act of writing, not before it.

2. Embrace Imperfection in First Drafts

Don’t aim for perfection in your initial drafts. The first draft is about getting ideas down; refinement comes later. As Anne Lamott famously said, "Almost all good writing begins with terrible first efforts."

3. Read Widely and Eclectically

While Lewis emphasizes reading "good books," don’t limit yourself to a single genre or style. Explore poetry, nonfiction, science fiction, historical accounts, and more. Each genre offers unique lessons in language, structure, and storytelling.

4. Observe the World Around You

Great writing often stems from keen observation. Pay attention to people, nature, conversations, and emotions. Keep a journal to jot down interesting details, phrases, or ideas that might inspire your work later.

5. Experiment with Different Forms and Styles

Try writing in different formats—short stories, essays, poetry, scripts, or even tweets. Experimenting with form can help you discover new ways to express ideas and stretch your creative muscles.

6. Seek Feedback, but Trust Your Instincts

Share your work with trusted readers or writing groups, but remember that not all feedback needs to be acted upon. Learn to distinguish between constructive criticism and opinions that don’t align with your vision.

7. Study the Craft of Writing

Read books on writing by authors like Stephen King (On Writing), Ursula K. Le Guin (Steering the Craft), or William Zinsser (On Writing Well). These resources can provide valuable insights into technique, voice, and structure.

8. Write for Yourself First

While it’s tempting to write for an audience or to chase trends, the most authentic and compelling writing often comes from writing what you care about. If you’re passionate about your subject, it will resonate with others.

9. Learn to Edit Ruthlessly

Good writing is often the result of rigorous editing. Be willing to cut sentences, paragraphs, or even entire sections that don’t serve the piece. As the saying goes, "Kill your darlings."

10. Cultivate Patience and Resilience

Writing is a long-term endeavor. Rejection, writer’s block, and self-doubt are part of the process. Learn to persevere and view challenges as opportunities to grow.

11. Find Your Unique Voice

Don’t try to imitate other writers too closely. Your voice—your unique way of seeing and expressing the world—is what will set your work apart. This develops over time through practice and self-reflection.

12. Write with Empathy

Whether you’re writing fiction or nonfiction, strive to understand and convey the perspectives of others. Empathy enriches your writing and helps you connect with readers on a deeper level.

13. Balance Solitude with Community

Writing is often a solitary activity, but connecting with other writers can provide support, inspiration, and accountability. Join writing groups, attend workshops, or participate in online communities.

14. Don’t Fear Failure

Every writer produces work that doesn’t meet their expectations. Failure is a natural part of the creative process. Learn from it, and keep moving forward.

15. Remember Why You Write

On days when writing feels like a struggle, reconnect with your purpose. Whether it’s to tell a story, explore an idea, or simply express yourself, remembering your "why" can reignite your passion.

C.S. Lewis’s advice provides a strong foundation, but writing is a deeply personal journey. The most important thing is to keep writing, keep learning, and stay true to your own voice and vision.

Now, this is even more useful for our present world. I told DeepSeek

one more thing, use GenAI like DeepSeek as a coach and tuitor

DeepSeek:

 Absolutely! Using generative AI like DeepSeek as a writing coach and tutor can be a game-changer for writers at any stage. Here’s how you can leverage AI to enhance your writing process:

1. Brainstorming and Idea Generation

Stuck for ideas? Use AI to generate prompts, plot points, or character concepts. For example, you can ask:

• "Give me 10 unique story ideas set in a futuristic city."
• "Suggest some themes for a short story about family secrets."

2. Overcoming Writer’s Block

AI can help you break through creative blocks by suggesting next steps or expanding on your ideas. For instance:

• "I’m stuck on this scene where two characters are arguing. Can you suggest some dialogue?"
• "Help me describe a mysterious forest in a way that feels eerie and magical."

3. Improving Clarity and Structure

Share your draft with the AI and ask for feedback on clarity, flow, or structure. For example:

• "Does this paragraph make sense? How can I make it clearer?"
• "Is the pacing of this story too slow? How can I improve it?"

4. Learning New Techniques

Use AI to learn about literary devices, narrative techniques, or genre-specific writing styles. Ask:

• "How can I use foreshadowing effectively in a mystery story?"
• "What are some ways to write a compelling opening sentence?"

5. Editing and Polishing

AI can act as a first-pass editor, helping you refine grammar, tone, and word choice. Try prompts like:

• "Can you help me make this sentence more concise?"
• "How can I make this dialogue sound more natural?"

6. Exploring Different Perspectives

AI can help you see your work from a reader’s perspective or suggest alternative approaches. For example:

• "How might a reader interpret this character’s actions?"
• "What’s another way I could end this story?"

7. Research Assistance

Use AI to quickly gather information or fact-check details for your writing. Ask:

• "What are some historical details about 19th-century London?"
• "What are the key characteristics of a dystopian society?"

---

8. Practicing Writing Exercises

AI can generate writing exercises to help you hone specific skills. For instance:

• "Give me a prompt to practice writing vivid descriptions."
• "Create a character backstory exercise for me to complete."

---

9. Experimenting with Style and Voice

Use AI to explore different writing styles or voices. For example:

• "Rewrite this paragraph in the style of Hemingway."
• "How would this scene sound if it were written in first person instead of third person?"

---

10. Setting Goals and Staying Accountable

AI can help you set writing goals and track your progress. Try prompts like:

• "Help me create a writing schedule for the next month."
• "What are some achievable goals for finishing my novel draft?"

---

11. Learning from Examples

Ask AI to analyze excerpts from famous works or provide examples of specific techniques.

• For example:

  "Show me how J.K. Rowling builds suspense in a scene."
• "What makes the opening of 1984 so powerful?"

---

12. Building Confidence

Use AI to validate your ideas or boost your confidence when you’re feeling unsure. For example:

• "Does this concept have potential for a novel?"
• "What’s working well in this piece so far?"

13. Exploring Genre-Specific Advice

If you’re writing in a specific genre, AI can provide tailored advice. Ask:

• "What are the key elements of a good fantasy world?"
• "How can I make my thriller more suspenseful?"

14. Collaborative Writing

Use AI as a co-writer to generate ideas, dialogue, or even entire scenes. For example:

• "Help me write a conversation between two characters who are meeting for the first time."
• "Suggest a twist ending for this story."

15. Reflecting on Your Work

AI can help you analyze your writing habits, strengths, and areas for improvement. Try prompts like:

• "What patterns do you notice in my writing style?"
• "How can I improve my descriptive writing?"

A Word of Caution:

• While AI is a powerful tool, it’s important to remember that it’s a supplement, not a replacement, for your creativity and judgment. Use it to inspire and guide you, but always stay true to your unique voice and vision.
• By treating AI like DeepSeek as a coach and tutor, you can accelerate your growth as a writer, experiment with new ideas, and refine your craft in ways that were never before possible. Happy writing! 🖋️

 

The recent DeepSeek Saga and Its Impacts

 The recent DeepSeek Saga.

DeepSeek capability, efficiency, and open-source licensing essentially break the profit-making dream of US AI companies like OpenAI (ChatGPT), Meta, Microsoft, etc.

Firstly, there are 83 hours of cyber attacks from the US on DeepSeek. China's privately owned internet companies such as 360 etc came together to fence off the attacks. It shows the unity of the people of the two countries and the bad behavior of the US.

OpenAI claimed but without proof that DeepSeek's efficiency comes from the 'distillation' of their work. Quite a number of people believed their claims. However, OpenAI and DeepSeek use different reasoning algorithms.

Architecture and Design

- Deepseek R1 uses a Sparse Mixture of Experts (MoE) approach with double reinforcement learning, reducing reliance on supervised training. It learns from its own mistakes, similar to xAI's Grok-1.

- In contrast, OpenAI o1 combines reinforcement learning with supervised training, often relying on models like GPT-4.

In layman's terms, it is like the proverb of work hard or work smart.

Output = work-hard x work-smart.

OpenAI is less smart so it needs to work harder, consuming a large amount of power for training and learning.

A very crude analogy is OpenAI is using Bubble sort whereas DeepSeek is using Tree-sort.

The great contribution of DeepSeek is that it reduces the costs of doing AI. It is open-source and efficient and allows countries without many resources to do the important AI work to build more effective and efficient processes and products. A big contribution to the well-being of the world.

What are your thoughts about the DeepSeek Saga?

Please share with me.

I feed my post to DeepSeek and ChatGPT, and here are their responses:

DeepSeek:

The "DeepSeek Saga" you describe is an interesting narrative that touches on several key themes in the AI industry, including technological innovation, competition, open-source versus proprietary models, and geopolitical dynamics. Here are some thoughts on the points you raised:

1. Cyber Attacks and Geopolitical Tensions

• The claim of 83 hours of cyber attacks from the US on DeepSeek, if substantiated, highlights the increasingly competitive and contentious nature of the AI race. It reflects how AI has become a strategic asset, with nations and corporations vying for dominance.
• The collaboration of Chinese internet companies to defend against these attacks underscores the importance of collective efforts in cybersecurity, especially in the face of sophisticated threats. However, it’s important to approach such claims with caution, as attributing cyber attacks to specific entities or nations can be complex and often requires concrete evidence.

2. OpenAI’s Claims and DeepSeek’s Innovation

• OpenAI’s suggestion that DeepSeek’s efficiency comes from "distillation" of their work is a common critique in the AI community, where advancements often build on prior research. However, as you pointed out, DeepSeek’s architecture (Sparse Mixture of Experts with double reinforcement learning) is distinct from OpenAI’s approach, which relies more on supervised training and reinforcement learning.
• The analogy of "working hard vs. working smart" is apt. DeepSeek’s approach seems to prioritize efficiency and adaptability, potentially reducing computational costs and energy consumption. This could be a significant advantage, especially as the environmental impact of large-scale AI training becomes a growing concern.

3. Architectural Differences

• DeepSeek’s use of Sparse Mixture of Experts (MoE) and double reinforcement learning is innovative. MoE models are designed to activate only relevant parts of the network for specific tasks, which can lead to faster and more efficient processing. This contrasts with OpenAI’s more traditional reliance on dense models like GPT-4, which require extensive computational resources.
• The comparison to sorting algorithms (Bubble sort vs. Tree-sort) is a good way to illustrate the difference in efficiency. DeepSeek’s approach appears to be more scalable and resource-efficient, which could democratize access to advanced AI capabilities.

4. Open-Source Contribution

• DeepSeek’s open-source model is a significant contribution to the AI community. By making their technology accessible, they enable smaller organizations and countries with limited resources to participate in AI development. This aligns with the broader movement toward open science and collaborative innovation.
• However, open-source models also come with challenges, such as ensuring responsible use and preventing misuse. Balancing openness with safeguards will be crucial for DeepSeek and similar initiatives.

5. Global Impact

• If DeepSeek’s technology lives up to its promise, it could disrupt the current AI landscape dominated by US-based companies like OpenAI, Meta, and Microsoft. By lowering the barriers to entry, DeepSeek could empower a more diverse range of stakeholders to innovate and compete.
• This could lead to a more balanced global AI ecosystem, reducing the concentration of power and fostering greater collaboration across borders.

6. Broader Implications

The DeepSeek Saga reflects the broader geopolitical competition in AI, where technological leadership is seen as a key determinant of economic and strategic power. The US-China rivalry in AI is likely to intensify, with both nations investing heavily in research and development.

At the same time, the saga highlights the importance of ethical considerations in AI development. As AI becomes more powerful, ensuring that it is used for the benefit of humanity will require international cooperation and shared standards.

Final Thoughts

The DeepSeek Saga is a fascinating case study in the evolving AI landscape. It showcases the potential for innovation to disrupt established players and the importance of open-source models in democratizing access to technology. However, it also underscores the need for transparency, collaboration, and ethical oversight as AI continues to advance.

If DeepSeek’s claims hold true, their approach could represent a significant leap forward in making AI more efficient, accessible, and sustainable. This would not only benefit the AI community but also contribute to solving global challenges in areas like healthcare, education, and climate change. At the same time, the geopolitical tensions surrounding AI development remind us that technology is never neutral—it is shaped by the values and priorities of those who create and control it.

ChatGPT: 

The DeepSeek Saga is fascinating and highlights several significant points in the evolving landscape of AI technology and geopolitics:

1. Innovation and Efficiency

DeepSeek's use of a Sparse Mixture of Experts (MoE) and double reinforcement learning is groundbreaking. The comparison to sorting algorithms (Bubble Sort vs. Tree Sort) is apt in illustrating how innovation can dramatically improve efficiency. If DeepSeek's architecture indeed allows it to "learn from its mistakes" autonomously and reduce training costs, this marks a pivotal shift in AI development.

2. Democratization of AI

The open-source nature of DeepSeek is a game-changer. By lowering the cost barrier, it enables smaller nations and companies to participate in AI innovation. This democratization is essential for fostering global development and ensuring that AI benefits everyone rather than being concentrated in a few wealthy corporations or countries.

3. Claims of "Distillation"

OpenAI's claims, whether valid or not, underscore a recurring tension in the tech world—how intellectual property, inspiration, and replication are defined and defended. If DeepSeek's architecture is truly independent, then accusations without evidence may appear as defensive posturing rather than legitimate concern.

4. Cyber Attacks and Geopolitics

The reported 83-hour cyberattack highlights the lengths some actors may go to suppress emerging technology that disrupts the status quo. The collaboration between private Chinese companies to fend off these attacks shows an inspiring level of unity and technical resilience.

5. Ethical and Economic Implications

DeepSeek's ability to reduce costs could shift the economic balance in AI, challenging the profit-driven models of US-based companies. While this may pose challenges for firms like OpenAI and Microsoft, it also forces them to innovate further, potentially leading to better AI products for everyone.

Final Thought

If DeepSeek continues to thrive and prove its claims, it could be a monumental contribution to the AI field and global technological progress. The Saga is a reminder of how open innovation, ethical practices, and resilience can reshape industries and defy monopolistic trends.

What’s your take on this, especially the cyberattack situation and OpenAI's claims?