Open models are decelerationist

Some guy called Dean Ball said

Open-weight models are inherently decelerationist

This other guy called Nathan Lambert says he’s right

Open models are decelerationist economically for the frontier labs, which will slow the net investment and capex rollout for AI. This is due to the fact that strong open-weight AI models massively reduce the margin potential for the closed labs. This has two effects. First, the AI labs have fewer profits to re-invest into future models. Second, the market sees the terminal value of these companies as being lower, so they will kneecap future fundraising rounds. These together will slow timelines to the most transformative AI models

I really want them both to be right, and they probably are since this outcome aligns neatly with China’s market-disruption strategy by way of open source commodification*¹.

To no one’s surprise except the valley-brained, open models are indeed the way. Just open-weight models for now, but that’s enough of a crack for the light to get in*².

The collective remit, then, is simple: If you’re using AI but you don’t think it should be centralized in the hands of a few intergalactic god-emperors, spend the bulk of your coin on open models and save the world.

Slow. It. Down.

It won’t slow to a grinding halt. Just enough that we can take some breaths, get regulated, and have the requisite spaciousness to deliberate together about what is next for us as a cybernetic species mid-evolutionary leap. We owe ourselves that much.