Isn’t the model fundamentally flawed if it can’t appropriately present arbitrary results? It is operating at a scale where human workers cannot catch every concerning result before users see them.
The ethical thing to do would be to discontinue this failed experiment. The way it presents results is demonstrably unsafe. It will continue to present satire and shitposts as suggested actions.
It won’t get people killed very often at all. Statistically there’s like no way you’ll know anybody who dies from taking a hallucinated suggestion. Give some thought to the investors who thought long and hard about how much money to put in. They worked hard and if a couple people a year have to die because of it how is that a bad trade off?
-kinda how it literally is almost unless the hubris is stronger than I imagine
Isn’t the model fundamentally flawed if it can’t appropriately present arbitrary results? It is operating at a scale where human workers cannot catch every concerning result before users see them.
The ethical thing to do would be to discontinue this failed experiment. The way it presents results is demonstrably unsafe. It will continue to present satire and shitposts as suggested actions.
It won’t get people killed very often at all. Statistically there’s like no way you’ll know anybody who dies from taking a hallucinated suggestion. Give some thought to the investors who thought long and hard about how much money to put in. They worked hard and if a couple people a year have to die because of it how is that a bad trade off?
-kinda how it literally is almost unless the hubris is stronger than I imagine