WhoBuilt

Why You Should Push Yourself to Do Hard Things

Jul 7, 2024· 50 min· 1.5K views·Watch on YouTube

Speakers

Nat Eliason, author, talks with Peter Yang about using AI in writing, building confidence through hard work, and creating lasting work without getting trapped by algorithms.

Use AI as an editor, not an authority

Nat uses AI tools such as Lex, Claude, and GPT mostly to critique nonfiction drafts, flag unclear or redundant passages, and check whether an argument treats opposing views fairly. He says the feedback can be specific, but writers need to judge it themselves: AI sometimes recommends a duller style that strips out personality. For fiction, he finds AI useful for research, brainstorming, and adding environmental detail.

Organize AI around a long project

While drafting his sci-fi novel Husk, Nat initially worked without AI, then used it to help manage a story with multiple viewpoints, many characters, and subplots. For a disease storyline, he used ChatGPT to brainstorm plausible diagnoses, and he used AI to explore names and character backgrounds. He also describes using NovelCrafter with a long prompt containing a writing sample and details about the characters, plot, and setting.

Build confidence with small wins

Nat says confidence in learning comes from repeated proof that you can figure things out. He suggests starting with a manageable task, such as fixing a washing machine, before taking on more difficult challenges. While writing Crypto Confidential, he tracked daily word counts in a spreadsheet; later, he built an app to log writing sessions and saw that he had written 110,000 words that year.

Choose work that lasts beyond the algorithm

Nat warns that creators can become dependent on algorithms and audience demand, repeating topics to keep views and income flowing. He argues that short-term engagement can distract from work with lasting value, such as books, and says creators should have a next step in mind when building an audience. For people who want to write books, he recommends starting a newsletter and considering a book deal once it reaches roughly 5,000 to 10,000 subscribers.

Takeaways

  • Use AI to find gaps and test arguments, then decide which suggestions fit your voice.
  • Break long-term work into visible inputs, such as daily word counts or logged work sessions.
  • Start learning with a small task you can complete to build evidence that you can teach yourself.
  • Build an audience around a durable body of work and decide what you want that audience to help you do next.

Mentioned in this talk

Search inside this talk

What was claimed 12

Related talks