> these labs are creating a public good as well.

https://en.wikipedia.org/wiki/Public_good

> In economics, a public good (also referred to as a social good or collective good)[1] is a commodity, product or service that is both non-excludable and non-rivalrous and which is typically provided by a government and paid for through taxation.

I can see that we are already excluding people from using models, whether it's China[1] or use in other harnesses[2] so its use can't be considered "non-excludable" or "non-rivalrous". It is also neither provided by the government nor funded through taxation.

I guess one change could be made to force LLM companies to release their (N-1)th model publicly and document architecture and system requirements, which would shift this to make these products actually public goods, however.

[1] https://x.com/AnthropicAI/status/2025997928242811253

[2] https://www.mindstudio.ai/blog/anthropic-openclaw-ban-oauth-...