In the open-source model of AI development, talented and expensive technologists who build AI models can still be paid for their work. However, the funding model would be different from the traditional approach of individual enterprises paying for the development of AI models.

One idea and probably the best, is for funding to come from government agencies or non-profit organizations that prioritize the development of AI for the greater good. For example, the US government's Defense Advanced Research Projects Agency (DARPA) has a long history of funding AI research, and non-profits like the OpenAI Foundation and the Partnership on AI also invest in AI development.

Another way is for companies to contribute to open-source AI development as part of their corporate social responsibility initiatives. This would allow them to support the development of AI that benefits society as a whole, while also promoting their brand as socially responsible.

People who work on open-source AI projects can also be paid through alternative funding models such as grants or crowdfunding. For example, the grant-making organization Mozilla has provided funding for AI development through their Mozilla Fellowship program.

The expensive part isn't the people, it's the mountain of GPU required to train the AI.