AI ALIGNMENT FORUMTags
AF

Mild Optimization

EditHistory
Discussion(0)
Help improve this page(2 flags)

Mild optimization is an approach for mitigating Goodhart's law in AI alignment. Instead of maximizing a fixed objective, the hope is that the agent pursues the goal in a "milder" fashion.

Further reading: Arbital page on Mild Optimization

Posts tagged Mild Optimization
Most Relevant
5
19When to use quantilization
Ryan Carey
2y
0
1
8Stable Pointers to Value III: Recursive Quantilization
Abram Demski
3y
0
1
8Quantilizers maximize expected utility subject to a conservative cost constraint
Jessica Taylor
5y
0
1
7Optimization Regularization through Time Penalty
Linda Linsefors
2y
2
1
3Quantilal control for finite MDPs
Vanessa Kosoy
3y
0
1
2Thoughts on Quantilizers
Stuart Armstrong
4y
0
Add Posts