It’s Time To Take A Break From ChatGPT (and its variants).

🌈 Abstract

The article discusses the environmental impact of the widespread use of ChatGPT and other generative AI models, highlighting their high electricity consumption, water usage, and carbon footprint. It also examines the broader implications of the growing reliance on AI technology in various sectors, including education and the workforce.

🙋 Q&A

[01] The Environmental Impact of ChatGPT

1. What are the key concerns raised about the environmental impact of ChatGPT?

• ChatGPT consumes over half a million kilowatts of electricity each day, and its electricity consumption is projected to rival entire nations by 2027.
• The AI technology behind ChatGPT requires powerful GPUs or TPUs that consume more power than standard CPUs, leading to high electricity usage.
• The cooling of the powerful supercomputers used by ChatGPT requires large amounts of water, with Microsoft using approximately 700,000 liters of fresh water during GTP-3 training sessions.
• ChatGPT has a high carbon footprint, generating 8.4 tonnes of CO2 annually, more than twice the emissions of an average person.

2. How does the environmental impact of ChatGPT compare to other AI-powered tools?

• Google's new AI feature, Google Overview, is also energy-intensive, with one search on Google's AI requiring 3 watt-hours of electricity, which is equivalent to charging 7.5 electric cars.
• The article suggests that the widespread industrial use of generative AI and the automatic addition of AI-powered features on websites are likely to have a greater environmental impact than individual users' interactions with ChatGPT.

3. What are the potential long-term consequences of the growing energy and water demands of AI technology?

• The article suggests that the exponential growth in AI electricity consumption, which is expected to increase 10-fold by 2026, could have significant environmental consequences, potentially rivaling the energy usage of entire nations.
• The high water usage required to cool the powerful supercomputers used by AI models is also a concern, with Microsoft discreetly taking 6% of the water used in the West de Moines district to power its data centers.

[02] The Broader Implications of AI Adoption

1. How does the use of AI technology impact the job market and workforce?

• The article suggests that most people's hopes for a decently paid job are dependent on their ability to use AI, with job listings mentioning AI advertising salaries 20% higher than those that do not.
• The UK has an AI workforce of 360,000, with AI contributing £3.7 billion to the economy, making it difficult for individuals to avoid using AI technology in the current job market.

2. What are the challenges in addressing the environmental impact of AI technology?

• The article suggests that it is not simple to stop using ChatGPT, as the widespread adoption of AI technology has become integral to various industries and the job market.
• The article also notes that big tech companies will continue to make updates and leverage new AI technology to stay ahead of the competition, making it difficult for individuals to boycott every tech company or go without technology in a tech-dependent world.

3. What are the potential solutions or actions suggested in the article?

• The article suggests that we can pressure tech companies to fully disclose their practices and not accept half-truths about their sustainability efforts.
• The article also emphasizes the importance of raising awareness about the environmental impact of AI technology and encouraging tech companies to prioritize sustainability in their development and deployment of AI models.