1 of 12

AI Safety:

How to help out

Toronto AI Safety Meetup

2024-10-31

2 of 12

It doesn’t have to be scary!

To work in AI Safety, you may need to:

  • Talk to people and network
  • Understand or solve difficult computer science issues
  • Navigate a confusing and still emerging field

This can seem daunting! But we’ll guide you through.

3 of 12

Other resources people recommended

Conferences - EAG (Effective Altruism Global, EAGx, Online)

Messaging people online - concrete question. LinkedIn

Maybe? Don’t worry about the Vase https://thezvi.substack.com

Import AI

Centre for AI Safety newsletter

Daniel Paleka

Waterloo - Cigi, rationality meetup

​

4 of 12

It’s a lot easier when you’re not working alone

Best ways to stay in touch if you’re in Toronto:

  • Meetup & Cafe Night every Thursday
    • Regulars are also invited to present on a safety-related topic of their choice�
  • AIGS Slack

​

UofT AI Safety Mailing List�

5 of 12

What kind of AI safetyist are you?

Policy and Governance

Technical Research

Present Day Harms

Catastrophic Future Risks

6 of 12

Ways to help

Learning

Networking

Research

Hackathons

Job opportunities

Volunteering

METR task bounties

YouTube

Alignment Forum

Read papers!

Meetups

Online

AIGS

Coworking space - Trajectory

This

Effective Altruism

Rationality

MLTO

80,000 Hours job board

Paper replications

Apart research

AIGS Slack

Courses

Neel Nandas 200 open problems

CognitionTO

AI Safety Camp

MATS

SPAR

7 of 12

Ecosystem - most of it not in Canada

  • A lot of AI safety orgs are based in hubs such as London (UK) and the Bay Area (California)
  • Some employ remotely
  • We’re building a community and infrastructure in Canada... but it takes time (and money)

Who funds all this?

  • Open Philanthropy (Dustin Moskovitz and others)
  • Survival and Flourishing Fund (Jaan Tallinn)
  • Private donors
  • Governments

8 of 12

AI Safety in Toronto

  • This meetup
    • Open to anyone
  • Trajectory office
    • People working independently on projects related to catastrophic risks from AI
    • Currently waiting to see if funding will be renewed
  • University of Toronto
  • Vector Institute

​

9 of 12

Kind of Roles, kind of a map of AI safety in general

AI Ethics

AI Governance

Technical AI Safety

  • Bias & Fairness
  • Privacy
  • Copyright issues
  • Plagiarism
  • Impact on employment
  • Environmental impact
  • Equitable access to AI
  • Transparency
  • Accountability
  • Disinformation
  • Preserving democracy
  • Impact on mental health
  • Corporate governance
  • State/province
  • National
  • International agreements

​

  • Laws and regulations
  • Voluntary agreements
  • Standards
  • Organizations and structures
  • Risk management/RSP
  • Public awareness�
  • Technical governance

​

  • Interpretability
  • Scalable oversight and control
  • Evals
  • Agentic AI
  • Theoretical - decision theory etc.

Support roles: operations, some software infrastructure, earning to give, etc.

10 of 12

Skilling up

What you already bring

Many kinds of skill will be useful for pursuing AI safety as a career or part-time interest.

​

​

Understanding AI safety

Get to learn the main arguments and a little about the ecosystem.

Develop deep expertise

You want to stand out as someone who really knows what you’re talking about.

Technical research, policy or support roles

Put what you’ve learned into practice

11 of 12

Non-Canada-specific resources

  • Courses
  • Online Communities
  • Projects
  • Jobs
  • Events
  • Funding and donation information

12 of 12

Remember: the AIGS slack is the best way to stay in touch!

#toronto-ai-safety-meetup

Meetup recordings, requests and banter

​

#study-buddies

Share interesting papers and report on your study progress

​

#introductions

Say hi to the Canadian AI safety community!

​

... and more.