Back to the MIT repository
6. Socioeconomic and Environmental2 - Post-deployment

Exclusion

The best AI techniques requires a large amount resources: data, computational power and human AI experts. There is a risk that AI will end up in the hands of a few players, and most will lose out on its benefits.

Source: MIT AI Risk Repositorymit97

ENTITY

1 - Human

INTENT

1 - Intentional

TIMING

2 - Post-deployment

Risk ID

mit97

Domain lineage

6. Socioeconomic and Environmental

262 mapped risks

6.1 > Power centralization and unfair distribution of benefits

Mitigation strategy

1. Prioritize and invest in Open Source AI initiatives to provide broader access to foundational models, data, and tools, thereby mitigating the concentration of technological control and fostering a competitive polyculture. 2. Develop and execute AI democratization policies focused on reducing computational and financial barriers to entry, ensuring that a wider range of individuals and enterprises can effectively utilize and benefit from AI capabilities. 3. Establish decentralized AI governance frameworks to distribute decision-making power and regulatory oversight across multiple public and private entities, preventing the concentration of undue influence.