Concrete actions anyone can take to help improve AI safety (with Kat Woods)

Jul 3, 2024 1h 9 insights Episode Page ↗
Spencer Greenberg speaks with Kat Woods, founder of Nonlinear, an AI safety charity, about the imperative to slow AI development due to existential risks. They discuss the difficulty of controlling superintelligent systems and what individuals can do to encourage safe AI.
Actionable Insights

1. Seriously Consider AI Risks

Actively reflect on the potential existential risks of AI, rather than dismissing or avoiding the topic. Acknowledge the high probability of severe outcomes and let this understanding motivate you to take action.

2. Prioritize AI Safety Testing

Treat AI development like medicine or food development, requiring stringent safety proof before release, rather than releasing and reacting to harm. This approach prioritizes safety over speed to prevent catastrophic outcomes.

3. Apply AI Golden Rule

Adopt a new Golden Rule: treat less intelligent beings (like animals) with the same care and consideration you would want a superintelligent AI to treat humanity. This fosters empathy and awareness of potential AI indifference towards us.

4. Acknowledge AI’s Alien Mind

Recognize that AI intelligence is fundamentally different from human intelligence, possessing superhuman capabilities in some areas and deficiencies in others. This understanding highlights the difficulty in predicting its behavior or truly aligning its motivations with human values.

5. Donate to AI Safety

Donate to AI safety organizations like Pause AI, Mana Fund (specifically regrantors like Dan Hendricks and Adam Gleave), or explore the Non-Linear Network to support work on slowing down AI development and ensuring its alignment with human values. Financial support is crucial for these efforts.

6. Volunteer for AI Safety

Volunteer your time and skills by joining the Pause AI Discord channel (found at pauseai.info/act) where numerous projects are listed, including opportunities for writing, research, development, legal advice, and petitioning. This directly contributes to AI safety efforts.

7. Contact Representatives on AI

Contact your political representatives by writing letters or making phone calls to express your concerns about AI safety and advocate for specific legislation. This direct communication is highly impactful as politicians pay attention to constituent feedback.

8. Online AI Safety Advocacy

Actively participate in online advocacy by loudly expressing concern about AI safety on social media, liking and sharing relevant posts, and using symbols like the ‘pause emoji’ in your profile. This raises awareness and signals public concern to politicians and corporations.

9. Support Specific AI Regulations

Advocate for specific AI regulations, such as limiting the computational power (compute) used for training frontier models or implementing remote shutdown capabilities in AI training hardware (GPUs). These measures can directly slow down and control dangerous AI development.