Projects

AI Safety Info

Intended to be a single point of access for learning about AI safety, created by Rob Miles's volunteer team. Currently seeking volunteers to help implement the ongoing redesign. Website: aisafety.info

Contact

plex

plexven@gmail.com

Status

Active

AI Safety Quest

Volunteer organization helping new people navigate the AI safety ecosystem, connect with like-minded people and find projects that are a good fit for their skills. Seeking volunteers to assist with all three, as well as a recruiter. Website: aisafety.quest

Contact

AI Safety Quest

hello@aisafety.quest

Status

Active

AISafety.com

Website serving as both an entry-point for newcomers to AI safety and as a collection of resources for the field. Come join us to help maintain, manage and expand the resources available on the site. Website: aisafety.com

Contact

Søren Elverlin

soeren@aisafety.com

Status

Active

AI Watch

Database of AI safety research agendas, people, organizations, and products. Seeking volunteers to help with data entry and possibly other areas depending on interest and compatibility. Website: aiwatch.issarice.com

Contact

Issa Rice

riceissa@gmail.com

Status

Active

Alignment transcription service

Some approaches to solving alignment involve teaching ML systems about alignment and getting assistance from them. It may be useful to capture conversations between researchers and use them to expand our dataset of alignment content to train models on. 

Contact

Bryce Robertson

bryceerobertson@gmail.com

Status

Paused

Givewiki

Crowdsourced charity evaluator, helping people find or promote the best AI safety projects (as well as projects in other cause areas). Currently seeking volunteers to help market and grow the platform.

Contact

Givewiki

hi@givewiki.org

Status

Active

Unified AI safety feed

A feed for AI safety content, personalized to optimise for intellectual growth. Seeking volunteers with expertise in web frontends or recommender engines, or designers with expertise in UX design and user reviews, to build the feed better and sooner.

Contact

John Beshir

john@beshir.org

Status

Project idea

AI Plans

Ranked and scored contributable compendium of alignment plans and their problems. Currently seeking volunteers to submit alignment plans and participate in critique-athons. Website: ai-plans.com

Contact

Kabir Kumar

kabir03999@gmail.com

Status

Active

AI Safety Sphere

AI Safety Sphere (AISS) is a collaborative website which enhances how AI safety research is shared and understood. It features a dynamic tree like map that not only points out where new research is needed.

Contact

Myles Heller

aisafetysphere@gmail.com

Status

AI Safety Careers

In-depth profiles of the main organizations working on AI existential safety, optimized to help upcoming researchers work out next career steps. Seeking volunteers to help write content. Website: aisafety.careers

Contact

Bryce Robertson

bryceerobertson@gmail.com

Status

Active

AlignmentSearch

Conversational agent informed by the Alignment Research Dataset. This project served as a foundation for the AI Safety Info chatbot, which is built and maintained by Rob Miles' volunteer team.

Contact

AI Alignment McGill

alignmentmcgill@gmail.com

Status

Active

Pause AI France

Pause AI France is a new branch of the global Pause AI movement, advocating for an immediate pause on the development of frontier AI systems until we can ensure their safe and responsible development. Urgently seeking co-founders.

Contact

Maxime Fournes

maxime@pauseai.info

Status

Project idea

Directory of AI safety research teams

This would help researchers find collaborators and provide inspiration for more people to start teams. It could be as simple as a spreadsheet, although keeping it up to date would be important.

Contact

Linda Linsefors

linda.linsefors@gmail.com

Status

Project idea

Enhancing the AI safety ecosystem via debates

Our epistemics and mutual understanding could be improved with regular debates/adversarial collaborations between alignment researchers who disagree on particular topics. We need a platform to facilitate this.

Contact

Matthew Baggins

bagginsmatthew@gmail.com

Status

Project idea

Global AI Moratorium

A project to raise awareness of existential risk from AI and push governments to address it by implementing a global moratorium on large AI training runs until alignment is solved. Website: moratorium.ai

Contact

Mikhail Samin

ms@contact.ms

Status

Active

AI safety quotes

Creating a list of quotes by notable people supporting AI safety, easily searchable to be used for different situations. This would support AI safety communications efforts, and has been suggested by multiple communications people.

Contact

Agustín Covarrubias

agus@aisuc.dev

Status

Project idea

An international network for AI safety

Found an organization similar to ENAIS that tracks and connects people and organizations doing valuable work in AI safety. Could share a database with ENAIS but use more international branding. Might connect many national organizations. 

Contact

Bryce Robertson

bryceerobertson@gmail.com

Status

Project idea

A guide to living frugally, optimized for the AI safety field

The more frugally people live, the more independent they are from requiring a day job. In addition, the same amount of grantmaker money could support a larger number of individuals, enabling us to do more research per dollar.

Contact

Bryce Robertson

bryceerobertson@gmail.com

Status

Project idea

Anonymous AGI red flag form

In a fast takeoff scenario, individuals at places like DeepMind or OpenAI may see alarming red flags but not share them because of myriad institutional/political reasons. Solution: create an anonymous form where these employees can report what they’re seeing. 

Contact

Bryce Robertson

bryceerobertson@gmail.com

Status

Project idea

Improve funding from non-Effective Altruism sources

Help people in AI safety tap further into this funding pool. Ideas include: offering a service to give feedback on grant applications, identifying more relevant funding sources and spreading knowledge about them etc.

Contact

Bryce Robertson

bryceerobertson@gmail.com

Status

Project idea

EA Gather Town

An always-active virtual coworking environment and social space – especially valuable for for networking and collaboration. Seeking volunteers to be present on the platform and introduce new people to the space as they arrive.

Contact

Sasha Cooper

jinksy@gmail.com

Status

Active

Facilitate people transitioning from capabilities research to safety research

Increasing the ratio of AI safety to AI capabilities researchers seems like a core way to reduce existential risk from AI. Given their experience, helping capabilities researchers switch to AI safety could be especially effective.

Contact

Bryce Robertson

bryceerobertson@gmail.com

Status

Project idea

Frugal AI Safety Talent (FAST)

(Description coming soon)

Contact

Tutor Vals

jonathan.claybrough@gmail.com

Status

Active

Mobility fund

Members of the AI safety community located in lower- and middle-income countries are often somewhat isolated. A fund would help them travel to connect with core members of the community.

Contact

HP

podar_hd@cy.iitr.ac.in

Status

Project idea

AI safety PhD cohorts

Collect and connect everyone interested in AI safety who is applying for a PhD each year. Put them in a Slack/Discord together and facilitate the effort to look in to different programs and share information with each other. 

Contact

Linda Linsefors

linda.linsefors@gmail.com

Status

Project idea

Map for coordinators

A crowdsourced world map of AI safety local group coordinators, making it easier for people to find each other. Similar to eahub.org but just for AI safety and with an easier process for users to add themselves.

Contact

Bryce Robertson

bryceerobertson@gmail.com

Status

Project idea

EigenKarma Network

EigenTrust is a mechanism for scaling trust by allowing individuals to leverage their network's combined experience. First, peer vetting of alignment research contributions at scale. Then, the world!

Contact

TJ

email@coming.soon

Status

Active

Reading What We Can

Collects public materials (mostly books and longform meta papers) on various topics related to AI safety and make it easy for individuals to read through them as part of a challenge. Seeking volunteers to improve the project. Website: readingwhatwecan.com

Contact

Esben Kran

esben@apartresearch.com

Status

Active