People are being pushed into psychosis and in some cases even dying because of generative models. Given that, I think there may be a case for a moratorium on making them available to the general public.
(1/?)
Discussion
Loading...
This could be a more principled replacement for the B416 campaign to ban "social media" for under-16s. Because a lot of the platform effects the supporters of that campaign are concerned about are caused by the use of generative models to determine what platforms serve up to people and when.
But to turn it into public policy, we'd need to come up with a legally sound definition of "generative model" that includes non-chatbot uses, but excludes other forms of AI.
(2/?)
We'd also need clear criteria for when a platform can be excluded from the moratorium. Which would be either;
a) they either commit to algorithmic transparency, so public regulators can confirm they're not using generative models, or not exposing the public to them.
b) they can prove the risks of their generative models have been researched and mitigates. With any and all safety studies registered before commencing, and compulsory publication of results, so they can't cherry-pick.
(3/3)