The Greatest Threat to Generative AI Is Humans Being Bad at Using It

I agree with all of this, especially the bits about a need for friction (which isnt going to happen in many spaces since AI apps are a race to the bottom), and the sudden change in public attitude.

I was in an ESRGAN img2img tuning community before the boom, and it was fun. Now... I would feel weird even disclosing any training runs or inference tests, like someone would PM me death threats or report me for trying to improve an old TV show or whatever.

When this will hit production systems and might have financial repercussions, then might need to prove wrong the label of nonsense talk like ChatGPT in politics and otherwise.

New technologies are created in the following order:

- Make it possible (Blackberry, ChatGPT)

- Make it pleasant (iPhone, ?)

- Make it cheap (Android, ?)

Generative AI is still in the 'Make it possible' stage.