MIT AI Risk Repository · Risk Sub-Category · 53.01.07
Language model misalignment
Category: Alignment failures in existing ML systems
Description
-
From Advancing AI Governance: A Literature Review of Problems, Options, and Proposals (Maas2023), as extracted by the MIT AI Risk Repository (CC BY 4.0).
Classification
Subdomain definition: 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.
Real-world incidents in this subdomain
- Reinforcement Learning Reward Functions in Video Games
- Predictive Policing Program by Florida Sheriff’s Office Allegedly Violated Residents’ Rights and Targeted Children of Vulnerable Groups
- Image Classification of Battle Tanks
How other frameworks describe this risk
- Natural Language Underspecifies Goals
- Loss of control
- Loss of control
- Sudden loss of control
- AI leads to humans losing control of the future
- Risks from delegating decision-making power to misaligned AIs
- Risks from AIs developing goals and values that are different from humans
- Future AI systems might actively reduce human control
Other entries from Maas2023
- Alignment failures in existing ML systems
- Faulty reward functions in the wild
- Specification gaming
- Reward model overoptimization
- Instrumental convergence
- Goal misgeneralization
- Inner misalignment
- Harms from increasingly agentic algorithmic systems
- Dangerous capabilities in AI systems
- Situational awareness
- Acquisition of a goal to harm society
- Acquisition of goals to seek power and control