Greener AI

When we think about AI, it gives people a more convenient life and a much more efficient and effective algorithm to produce the ‘AI generated’ result. However, if we ask if AI makes the greener world? The answer might not be ‘yes’.

Then the ‘Green AI’ concept comes out. Green AI, also known as Sustainable AI or Eco-friendly AI, refers to the development and deployment of artificial intelligence technologies that minimize their environmental impact and promote sustainability.

For instance, it is estimated that only the training of the popular GPT-3 model, a 175-billion parameter LLM, emitted approximately 502 tonnes of CO2.

https://appen.com/blog/smart-solutions-for-a-greener-future-how-ai-is-making-a-difference/

How we can make green AI? There are two categories in making greener AI.

• Hardware
• Software

Hardware

1. Hardware Design: Designing energy-efficient hardware specifically tailored for AI tasks can significantly reduce power consumption. Specialized AI accelerators, such as graphics processing units (GPUs) or tensor processing units (TPUs), are designed to perform AI computations more efficiently. Additionally, exploring emerging technologies like neuromorphic computing or photonic computing can potentially lead to further energy savings. IBM brain-like chip (Eco-friendly AI: IBM’s brain-like chip prototype will be a massive breakthrough for green LLMs.)