Mastodon Feed
Reblogged by collinsworth@hachyderm.io ("Josh Collinsworth"):
sysop408@sfba.social ("Sheldon Chang 🇺🇸") wrote:
This whole ordeal with #OpenAI tracks against my past dealings with AI. You could get the AI 90% of the way to doing the job you wanted it to do, but attempts to close the last 10% would result in the project moving backward.
Because of this, we could never take a human off of closely babysitting the AI's decisions, which didn't actually save us much time. It just changed the nature of the work for the human.
If we went all in and and tried to close the final 10% to make the AI self-sufficient, the whole thing would go haywire to the point of being unusable.
So at least for me, OpenAI blowing up as soon as they could see the promised land is totally on brand. 😂