Introduction to AI 2

by Albert Major

Contemporary AI issues

1. AI’s Energy Footprint

Context

• AI models consume a significant amount of electricity and water.
• Example: Wildfires in Los Angeles require a lot of water, but so do data centers in Silicon Valley, which need cooling.
• Data centers contribute to water shortages and climate concerns.

Key Issue

• Should AI models have energy ratings like household appliances (e.g., A, B, C efficiency labels)?
• AI models consume varying amounts of energy for different tasks.
  • Example: A thousand text classification tasks use as much energy as running a laptop for 10 hours.
  • Example: Image generation consumes significantly more energy.

Proposed Solutions

• Regulation of AI’s environmental impact.
• Industry self-regulation (companies like Google promise carbon-free energy but exclude AI operations).
• Debate on whether AI’s long-term benefits outweigh its current environmental costs.

Discussion Question:

Do we need regulations to prevent AI models from causing harm to the environment?

• Pro-Regulation Argument:
  • AI models consume vast amounts of energy and water.
  • Without regulation, companies may not act responsibly.
  • Policies could ensure sustainable AI development.
• Anti-Regulation Argument:
  • Companies already have financial incentives to optimize energy use.
  • Regulations could stifle innovation and favor big corporations.
  • Startups would struggle to meet compliance costs.