Welcome
Thank you all for signing up to the first DC<>AI Bootcamp. Over the next week you will be fully immersed in the cutting edge of AI, taught directly by the various stakeholders and factions of San Francisco and the East Bay. I’d like to extend a special welcome to Senator Sanders and Senator Cotton for their attendance; we understand you’ll be coming in and out of the room with your other responsibilities but we value very much your presence.
We’ll begin with your setup. You have in front of you a MacBook Pro loaded with Claude, ChatGPT and Gemini, with the newest models working in their web products and directly in the CLI, with a fully-loaded Obsidian back-end that works between these three platforms. This approach to work may feel unfamiliar to you but the basis of the idea here is that your computer has only one button that matters to you: the left-hand function key to open up Wispr Flow: you use the AI and the AI uses the computer.
You’ll hear from a series of technical leaders over the first two days, encouraging you to build up your competency using these tools and push you out of your comfort zone of the keyboard and into a brave new world.
From there we’ll enter into two days of faction overviews, beginning with the most extreme examples of accelerationism hosted by Beff Jezos, to the most extreme side of AI safety with a presentation from MIRI. From there we’ll fill in the middle factions: folks that are worried about safety from the perspective of mechanistic interpretability and alignment and those who believe these are ordinary tools and everything you’ve heard about AI safety is a bunch of hype.
We’ll have every day begin with lectures after breakfast, followed by a period of one-on-one coaching in the afternoon. This cohort of 100 represents staffers from across both the House and the Senate, as well as analysts from all of the major foreign and domestic think tanks who are excited to jump from novice to expert.
On the last two days you’ll be working with your mentors one-on-one to produce a poster similar to what would be displayed at AI conferences. On this poster you will be working directly with the AI of your choice to produce a policy document, collaborating with this AI to show its impact on your specific sector- could be agriculture, could be foreign policy, could be cyber defense. You’ll be presenting this work to your boss when he arrives on Tuesday for a series of “best-of” lectures and then a showcase of your work. This day is deliberately designed to give your immediate supervisor a taste of the experience you’ve just gone through, while also showcasing how AI can be used to make your work more appealing and more streamlined.
It’s our objective to get you settled in and comfortable with Frontier AI tools- and to have you delight in their use to make your work better. Along the way you’ll learn about safety, you’ll learn about accelerationism, and you’ll learn about the future that is being built in San Francisco.
Some things you learn this week may be scary. Some things may be very optimistic. You’ll leave with a comprehensive understanding of what Silicon Valley is building and why these things matter to government. You will be able to spread the messages and knowledge of this time into your community here in Washington D.C., empowering our government to do better work in the future while working alongside AI and understanding the risks of this technology.
The idea is simply that the people who help write American AI policy should spend a week actually using these tools, meeting the factions and making something of their own before deciding what any of it means for government.
Pay attention, drink your coffee, and godspeed to cohort one of the DC<>AI Bootcamp.
Make It Real
$350k and a connection to a Uni in DC that can host, from the 6th-12th of January. Want to run it in Ottawa or Paris too?
sources & notes (AI generated)
Senators Sanders and Cotton — proposed attendees in this imagined welcome address, not confirmed participants.
Beff Jezos — pseudonym of Guillaume Verdon, physicist, Extropic founder and co-founder of effective accelerationism (e/acc); identity reported by Forbes in December 2023. Wikipedia. https://en.wikipedia.org/wiki/Guillaume_Verdon
MIRI — Machine Intelligence Research Institute, Berkeley nonprofit founded in 2000, since 2024 focused on halting development of increasingly general AI through policy work. About page. https://intelligence.org/about/
Wispr Flow — voice dictation app for Mac, Windows, iOS and Android; the Mac default hotkey is hold or double-tap the fn key. Help center. https://docs.wisprflow.ai/articles/2772472373-what-is-flow
Claude, ChatGPT and Gemini in the CLI — Claude Code (Anthropic), Codex CLI (OpenAI) and Gemini CLI (Google) are the three labs’ terminal agents. https://code.claude.com/docs · https://learn.chatgpt.com/docs/codex/cli · https://github.com/google-gemini/gemini-cli
The bootcamp itself — a proposal by the author, not an existing program; the “Make It Real” line is the ask: $350k, a DC university host, 6-12 January 2027.
What I love about this post is how directly it calls out a huge, frustrating problem in modern society.
It’s way bigger than just those awkward Congressional hearings where politicians act like grandparents asking their grandkid how to connect to Wi-Fi. Of course, it’s not really about the age gap, but the fact that they are just way too far from the technological front lines.
It gets straight to what scholars like Sowell and Taleb have warned us about: the people making the big decisions are way too far from the real world, and they never have to pay the price for getting things wrong.
Classical military generals and modern business leaders all know a simple truth: the people who can actually hear the gunfire should be the ones calling the shots. This post is all about bringing that exact rule back.