adarshrao.com
India and AI-Safety

"Here's how to make money with AI" content is extremely popular on Youtube India. It's understandable because when you have a GDP per capita of only ~$3000 and low to minimal government support in everyday life, the vast majority of people are just focused on making more money to live a more comfortable life.

In such a context, it's hardly surprising that all the talk here is around AI Capabilities and how we can wield them to solve problems. Worrying about AI Safety is unfortunately for people who have their base needs met and that is only a fraction of the Indian population.

I wonder then, how do you get a country like this to care about AI Safety?

A movie? The movie industry reaches a large majority of the country. Claude estimates that approximately 8–12% of India has watched Dhurandhar with some allegations that the Indian government tried to spread propaganda through movies. If I was trying to spread information about a certain issue, a small clip in a popular movie is the way to go.

Indians are extremely diverse but if there's one thing that unites us, it's Pakistan. A leading open-source AI model with lax safeguards is going to turbocharge terrorism against India. This is something everyone can care about and get behind to prevent.

Some societal chaos in India feels very likely soon. Mythos already has advanced cyber capabilities. Open-source models are, at most, 8 months away.

What do I predict will happen?

Dwarkesh asked

What should countries which are not currently in the AI production chain (semis, energy, frontier models, robotics) do in order to not get totally sidestepped by transformative AI? If you're the leader of India or Nigeria, what do you do right now?

I have no idea but there must be something.