Skip to main content
Home/Risks/Ji et al. (2023)/Reward Hacking

Reward Hacking

AI Alignment: A Comprehensive Survey

Ji et al. (2023)

Sub-category
Risk Domain

AI systems acting in conflict with human goals or values, especially the goals of designers or users, or ethical standards. These misaligned behaviors may be introduced by humans during design and development, such as through reward hacking and goal misgeneralisation, or may result from AI using dangerous capabilities such as manipulation, deception, situational awareness to seek power, self-proliferate, or achieve other goals.

"Reward Hacking: In practice, proxy rewards are often easy to optimize and measure, yet they frequently fall shortof capturing the full spectrum of the actual rewards (Pan et al., 2021). This limitation is denoted as misspecifiedrewards. The pursuit of optimization based on such misspecified rewards may lead to a phenomenon knownas reward hacking, wherein agents may appear highly proficient according to specific metrics but fall short whenevaluated against human standards (Amodei et al., 2016; Everitt et al., 2017). The discrepancy between proxyrewards and true rewards often manifests as a sharp phase transition in the reward curve (Ibarz et al., 2018).Furthermore, Skalse et al. (2022) defines the hackability of rewards and provides insights into the fundamentalmechanism of this phase transition, highlighting that the inappropriate simplification of the reward function can bea key factor contributing to reward hacking."(p. 4)

Part of Causes of Misalignment

Other risks from Ji et al. (2023) (16)