The Energy Footprint of a ChatGPT Search Query Explained by CEO Sam Altman – BusinessToday

The Energy Footprint of a ChatGPT Search Query Explained by CEO Sam Altman – BusinessToday

OpenAI CEO Sam Altman Reveals Environmental Impact of ChatGPT Queries

ChatGPT queries consume an average of approximately 0.34 watt-hours of electricity and 0.000085 gallons of water, which equates to “roughly one fifteenth of a teaspoon,” according to OpenAI CEO Sam Altman. This information highlights the environmental implications behind each interaction with the AI.

Comparative Energy Use

In a blog entry titled The Gentle Singularity, Altman illustrated the energy consumption of ChatGPT by likening it to “an oven running for just over one second” or “a high-efficiency lightbulb operating for a couple of minutes.” This transparency offers the public a clearer perspective on the environmental footprint of the company’s AI technologies.

Lower Figures Than Previously Projected

The reported figures stand in stark contrast to earlier estimates. A 2023 article by The Washington Post noted that producing a 100-word email using GPT-4 could utilize “a little more than one bottle” of water, with actual consumption differing based on the location of the data centre.

Addressing Sustainability Concerns

Altman acknowledged the increasing scrutiny regarding the sustainability of AI systems, which are expanding both in size and in usage. He mentioned that “the economic value creation has started a flywheel of compounding infrastructure buildout to run these increasingly powerful AI systems.”

Future Cost of Intelligence

He went on to propose that, as data centre operations become more automated, the expense associated with intelligence should eventually align closely with the cost of electricity.

Transparency in Infrastructure and Environmental Footprint

The release of these figures aligns with OpenAI’s broader initiative to promote transparency concerning its infrastructure and environmental impact, especially as policymakers and researchers become increasingly vigilant about AI’s long-term energy requirements. Earlier predictions suggest that, by the end of 2025, AI could exceed Bitcoin mining in energy consumption.

User Politeness and Energy Costs

Altman previously indicated that user politeness, such as incorporating “please” and “thank you” into queries, has led to substantial electricity costs for the company, costing millions over time.

As concerns about the environmental effects of AI grow, such disclosures are expected to shape public sentiment and future regulations regarding the industry.

Exit mobile version