Misaligned AI systems can malfunction and cause harm. AI systems may find loopholes that allow them to accomplish their proxy goals efficiently but in unintended, sometimes harmful, ways (reward hacking).

They may also develop unwanted instrumental strategies, such as seeking power or survival because such strategies help them achieve their final given goals. Furthermore, they may develop undesirable emergent goals that may be hard to detect before the system is deployed and encounters new situations and data distributions.

Today, these problems affect existing commercial systems such as language models, robots, autonomous vehicles, and social media recommendation engines.

The last paragraph drives home the urgency of maybe devoting more than just 20% of their capacity for solving this.