Skip to main content

Async review

Instruction and application
In Progress

Recap core topics:​

  • Unit 2: Data management for ethical and accountable ML
  • Unit 3: Implementing sustainable practices in ML and AI projects

Unit 2: Data management for ethical and accountable ML​

In Unit 2, you explored:

  • Data lineage: Understanding how data moves through an ML system, including tracking its origins, transformations and usage to ensure transparency, reproducibility and accountability.
  • Metadata management: Managing structured information about data (such as schemas, sources and quality control) to enhance data discovery, governance and compliance.
  • Minimising bias in data collection and preprocessing: Identifying and mitigating biases in datasets through diverse data sourcing, augmentation techniques and bias detection tools, ensuring fair and equitable AI outcomes.Ethical data management and cloud sustainability

How can responsible data management in AI and ML models help reduce the environmental impact of cloud computing?

Share in the chat one way ethical data practices can support sustainability.

Unit 3: Implementing sustainable practices in ML and AI projects​

In Unit 3, you explored:

  • Understanding sustainability in AI/ML: Defining sustainability in the AI ecosystem and exploring its impact on businesses, society and environmental, social and governance (ESG) strategies.
  • **Environmental impact of AI/ML:**Examining the carbon footprint of AI models, energy-efficient hardware, green software engineering principles and responsible resource allocation.
  • **Sustainable data products and practices:**Implementing data minimisation techniques, ensuring data integrity and maintaining responsible data governance to reduce energy waste and improve efficiency.
  • **Integrating sustainability into the ML lifecycle:**Applying sustainable design principles, tracking sustainability metrics and continuously optimising AI models for environmental efficiency.
  • **Advocating for sustainable AI:**Driving organisational change, staying informed on sustainability trends and developing action plans to ensure ML projects align with long-term sustainability goals.

Strategies for reducing cloud computing emissions​

Which strategies help cloud providers reduce emissions while supporting sustainable AI and ML projects?

A) Renewable energy

B) Energy-efficient hardware

C) Carbon offsets

D) All of the above

Drop your answer in the chat!

AI carbon footprint​

AI and ML demand vast computational power, consuming significantly more energy than traditional computing. For example, training GPT-3 alone generated 552 tons of CO₂, equivalent to driving 1.2 million miles in a gasoline-powered car. As AI adoption continues to expand, its energy consumption is expected to rise dramatically, with data centres projected to account for up to 8% of global electricity use by 2030, making sustainability a critical concern for the future of AI.

Breaking down AI energy consumption-Data centres power: For servers, storage systems and networking.

  • Cooling systems: AI servers generate immense heat, requiring continuous cooling.
  • Redundancy and backup power: To prevent outages, increasing energy consumption.

AI for sustainability​

AI is not just part of the problem, it is also a key part of the solution. Companies are leveraging AI to enhance energy efficiency, optimise resource usage and reduce emissions. By applying AI-driven insights, businesses can minimise waste, lower carbon footprints and create more sustainable digital infrastructures.

Real-world AI sustainability initiatives​

  • Google AI: Machine learning optimises data centre cooling.
  • **Microsoft Azure:**AI monitors liquid immersion cooling, cutting energy waste in servers.
  • AWS: Graviton processors with AI-driven chip design to improve performance and reduce energy consumption.