Project AI Kavach : AI safety for builders fellowship

This AI safety fellowship, dubbed Project AI Kavach, is focused on builders. Think of it like SPAR or MATS or ERA, but with mentors from the industry and mentees who are software engineers, product builders, etc. Anyone who can build industry-grade products, tools, or artifacts on AI safety problems.

Why? one thing I observed was that there is lot of research & empirical work happening in AI safety, but much less focus on building products and tools.

This gap is increasingly being recognized by several organizations.

• Organizations like MATS and BlueDot have been coming up with field building initiatives.
• Surplus.dev is like YC for AI safety & adjacent problems
• Forethought has recently released an RFP seeking useful AI safety tools
• Apollo wrote about the case for AGI safety products
• Halcyon has an RFP for safety-tooling founders

I think this reduce x-risk from AI in at least two ways:

1. Bring AI safety work from research orgs / toy implementation to real world faster. For example while there's work on human-AI interaction harms, and AI assisted biorisk - the implementations are either lacking or only research grade, rather than product grade.
2. It is also a way to engage population & ecosystems which are traditionally more focussed / interested in building than researching (for example Bengaluru, SF Bay Area, etc). AI safety for builders will mobilize / evangelize great builders to AI safety concerns (who may either be unaware or hijacked by the Tech Right (think a16z)

• Salaries, wages : 25,000 $ (1 quarter), 100,000 $ (1 year)
• Ops & marketing : 10,000 $ (1 quarter), 40,000 $ (1 year)
• Compute : 10,000 $ (1 quarter), 40,000 $ (1 year)
• Mentee support: 5,000 $ (1 quarter), 20,000 $ (1 year)