Generative models, especially those that use deep learning techniques, require vast amounts of resources to train, test, and deploy. Training a model can take days or weeks. This process requires powerful processors that consume large amounts of electricity and produce significant greenhouse emissions. The hardware that runs AI models – primarily GPUs – often contains rare metals (e.g., nickel, cobalt, and lithium) that are costly and environmentally taxing to collect and process. Data centers that house models generate significant heat and require substantial water and energy to cool. Secondary environmental impacts include emissions from AI-enabled applications. The resource requirements of AIs can impose significant costs on the natural environment, as they are often acquired and used in ways that are unsustainable, deplete resources, and damage built environments.
Excerpt from the MIT AI Risk Repository full report
The development and operation of AI systems causing environmental harm, such as through energy consumption of data centers, or material and carbon footprints associated with AI hardware.
Incident volume relative to governance coverage — each dot is one of 24 subdomains
Entity
Who or what caused the harm
Intent
Whether the harm was intentional or accidental
Timing
Whether the risk is pre- or post-deployment
Elon Musk's AI company xAI established a massive supercomputer facility called 'Colossus' in Memphis, Tennessee, operating up to 35 unpermitted methane gas generators that significantly increased air pollution in a predominantly Black, low-income community already burdened with industrial pollution and elevated health risks.
Developers: Xai
Deployers: Xai
Vulnerabilities that can be exploited in AI systems, software development toolchains, and hardware, resulting in unauthorized access, data and privacy breaches, or system manipulation causing unsafe outputs or behavior.
66 shared governance docs
AI systems that fail to perform reliably or effectively under varying conditions, exposing them to errors and failures that can have significant consequences, especially in critical applications or areas that require moral reasoning.
59 shared governance docs
AI systems that memorize and leak sensitive personal data or infer private information about individuals without their consent. Unexpected or unauthorized sharing of data and information can compromise user expectation of privacy, assist identity theft, or cause loss of confidential intellectual property.
54 shared governance docs
Challenges in understanding or explaining the decision-making processes of AI systems, which can lead to mistrust, difficulty in enforcing compliance standards or holding relevant actors accountable for harms, and the inability to identify and correct errors.
52 shared governance docs
Facilitates AI infrastructure development by easing federal regulations, utilizing federal lands for data centers, and providing financial support for AI projects. Directs various agencies to expedite environmental reviews, permitting, and financial assistance for AI-related projects.
Requires federal agencies to identify, solicit proposals for, and lease federal lands suitable for AI data centers and clean energy projects to support centers. Requires a plan to coordinate AI infrastructure development with U.S. allies including workforce development.
Guides AI developers and users in California on compliance with existing laws governing consumer protections, data protections, civil rights protections, competition laws, and for new AI-specific laws.