CIP Q2: Open-sourcing CCAI, hiring a founding engineer, and launching global inputs!

It’s been another big quarter at CIP: we’re hiring for a founding engineer, launching a process for global inputs into AI policy (already some fun responses in pilots), and we’ve released our roadmap for the next few months of prototyping. And we have a new office in Japantown! More on all of this and more below. As always, reach out at hi@cip.org if you’re interested in collaborating.

Join CIP as our Founding Engineer

Want to help us build collective intelligence? We’re hiring for a Founding Engineer role to take the lead in shipping technical prototypes for the democratic governance of AI.

If you’re entrepreneurial, mission-aligned, and excited about the possibility of building a team and shaping our technical roadmap (or know someone who is), check out the job posting and apply at: https://cip.org/careers/foundingengineer

Help us reach our $500,000 matching grant

We currently have a matching opportunity of up to $500,000 (for money raised by EoY) from a generous donor, and would be honored if you would consider helping us reach this goal. As a small non-profit funded by philanthropic grants from foundations and individuals, this would be a transformative amount for us.

While every dollar helps, just as important is spreading the word or connecting us with people you think we should meet.

https://cip.org/donate

We also receive non-cash donations; if you’d like to discuss that option please reach out to us.

Building open-source CCAI (and a few other things)

CIP is doubling down on building tools to enable better, more well-informed collective input into important decisions, especially decisions about governing technology.. Our Collective Constitutional AI work with Anthropic demonstrated that public input can improve foundation models. Now we are working on open-sourcing that CCAI project, with the goal of enabling any community to build a model aligned to their own preferences, and better understand collective values around technology along the way. 

We believe this is a fundamental way to democratize technology and empower communities. Open source CCAI is a fundamental first step in building towards our other collective intelligence tools this year. We are running a few other public input pilots in parallel, including partnering with Remesh to build a process for global input into AI. 

Get in contact if you’d like to work with us on this (on building this or using it), and especially if you see yourself as a potential founding engineer – or if you want to contribute a constitution co-written by your community!

CIP at FAccT in Brazil

Divya and Saffron presented our Collective Constitutional AI Project with Anthropic at the FAccT Conference in Rio De Janeiro. The paper and presentation details the work we did to gather collective input and values from a representative sample of the United States, and retrain Anthropic’s Claude Model on that data. What we found is that, using this process, you can create a less biased and better performing model more reflective of broader values. Read the paper at https://arxiv.org/abs/2406.07814

More recent publications:

We Need Network Societies, Not Network States

While the nation-state as the primary vehicle of welfare and governance continues to show it’s fault lines in the 21st century, it’s clear that we need to rethink the institutions that define our modern states. Co-published with Glen Weyl and Anne-Marie Slaughter, we revisit Balaji Srinivasan’s book “The Network State,” calling into question his technopolitical project and advocating for a model of governance that harnesses networks through the open-ended relationality of societies, not the closed-borders of nations.

Beyond Public and Private: Collective Provision Under Conditions of Supermodularity

The unique qualities of transformative technologies such as artificial intelligence transcends classical economic framings of public vs. private, rivalrous vs. anti-rivalrous, and demand new economic narratives. Along with Matthew Prewitt and Glen Weyl, we introduce the concept of a supermodular good, in which the value scales with each user added, but is also vulnerable to private capture and monopoly at different stages of the supply chain. We call to “make public provisioning systems more decentralized (addressing legitimate critiques of central, state-led provision) without sacrificing public benefits or shared ownership.”

New CIP staff members, transitions, and an office!

In the past three months we’ve onboarded Evan Hadfield (Program Lead), Joal Stein (Comms and Ops Lead), Zarinah Agnew (Research Director), and Audrey Tang (Senior Research Fellow). Check out their bios on our website at cip.org/about. 

Two years after co-founding CIP with Divya, Saffron will be moving to Anthropic to work on the societal impacts of frontier AI, and continue our shared work on collectively-steered models from the Anthropic side. We’re so proud of her! Saffron has brought so much insight, care, and joy to our work so far— spearheading our CCAI collaboration, leading societal impact evaluations in our partnership with UK’s AISI, org-building, and always bringing little treats to meetings. She’ll continue to play a key role within CIP: onboarding our new Founding Engineer, running our tech advisory board, and coworking out of our new office :) 

Because yes - after two years as a remote organization, we’ve moved into our first office in San Francisco in Japantown! Reach out to us if you’d like to come visit, co-work, or grab matcha.