Prompt Science

AI, research, and the frontiers of discovery — a community-audited podcast.

RSS Feed GitHub 𝕏 / Twitter ▶︎ Listen

Episodes

New episodes weekly. Audio files are hosted in /episodes. This list can auto-load from episodes/episodes.json.

Subscribe

https://github.com/napoles-uach/promptscience/feeds/podcast.rss Open RSS

Use this RSS in your favorite podcast app. Apple/Spotify links can be added once directories approve the show.

Spotify Apple Podcasts

What is Prompt Science?

Prompt Science is a podcast where large language models help draft scripts from recent, high-quality sources, and the community reviews, corrects, and enriches them. Every episode documents the models used, prompts, and references, so the process is transparent and reproducible.

Episodes are generated with AI, reviewed by humans, and published with sources, transcripts, and version history. When the script updates, the audio is regenerated and a new version is published.

Typical episode pipeline
  • Collect recent items from top venues (Nature, Science, arXiv, institutional newsrooms).
  • Draft a script with LLMs (tone: clear, curious, neutral American narration).
  • Human review: fact-check, adjust dates, add author names and citations.
  • Text-to-speech renders the final audio; MP3 is stored in /episodes.
  • Publish to GitHub Pages, update RSS, announce.

How it works

Generation: LLMs draft the episode from recent sources with explicit prompts and guardrails.

Verification: Contributors open Pull Requests to correct facts, add context, or improve clarity. All changes are version-controlled.

Reproduction: We track model names, versions, and prompts so episodes can be regenerated deterministically where possible.

Files and structure 
episodes/
  2025-10-05-navier-stokes.mp3
  2025-10-05-navier-stokes.md     ← transcript
  episodes.json                   ← optional index
assets/
  cover.png
feeds/
  podcast.rss                     ← your RSS feed
index.html

Contribute

Spotted an error? Want to add a paper or clarify a claim? Contribute on GitHub with a Pull Request. Edit the transcript Markdown, update references, or add a new episode card to episodes/episodes.json.

Open GitHub Repository
How to add a new episode
  1. Add the MP3 to /episodes.
  2. Add a transcript .md with references.
  3. Append an entry to episodes/episodes.json with title, date, duration, and file paths.
  4. Open a Pull Request. A maintainer merges after quick checks.

Submit corrections

Prefer not to use GitHub? Send corrections or sources by email. Include the episode title, the exact sentence, and a reliable citation.

📬 scienceprompt@gmail.com

Minimal checklist for corrections
  • Concrete reference (DOI, arXiv, Nature/Science URL).
  • Accurate dates and author names.
  • Direct quote or paraphrase with citation.
  • If adding numbers, show how they were computed.

Contact

📬 Email: scienceprompt@gmail.com

We’re preparing Apple/Spotify listings; links will appear here once approved.