Report on ChatGPT User Growth and Environmental Impact in the Context of Sustainable Development Goals (SDGs)

Introduction: Innovation and Responsibility

A recent confirmation from OpenAI indicates that the ChatGPT platform has reached 700 million weekly active users, a significant increase from 500 million in March and a fourfold expansion over the past year. This growth, spanning free, educational, and enterprise tiers, highlights the rapid integration of Artificial Intelligence into society. However, this technological advancement presents substantial challenges to global sustainability efforts, particularly concerning the United Nations Sustainable Development Goals (SDGs). This report analyzes the platform’s growth and its environmental footprint, assessing its alignment with key SDGs, including SDG 6 (Clean Water and Sanitation), SDG 7 (Affordable and Clean Energy), SDG 9 (Industry, Innovation, and Infrastructure), SDG 12 (Responsible Consumption and Production), and SDG 13 (Climate Action).

User Growth Analysis and Linkage to SDGs

Metrics of Expansion

The platform’s user base has demonstrated exponential growth, indicating widespread adoption across multiple sectors. This expansion is a testament to AI’s increasing role in global innovation.

• Weekly Active Users: 700 million
• Annual Growth: Quadrupled within the last year
• User Segments: Individuals, educational institutions, and businesses (Free, Plus, Pro, Enterprise, Team plans)

Implications for SDG 4 and SDG 8

The diverse adoption of ChatGPT directly impacts several SDGs. The integration into educational plans supports SDG 4 (Quality Education) by providing new learning tools. Simultaneously, its use in enterprise settings contributes to SDG 8 (Decent Work and Economic Growth) by driving innovation and productivity. However, this growth must be managed sustainably to ensure it aligns with broader environmental goals.

Environmental Footprint Assessment

The infrastructure required to support 700 million weekly users has a significant environmental cost, directly challenging the achievement of several critical SDGs.

Energy Consumption and SDG 7 (Affordable and Clean Energy)

The high energy demand of AI models and their supporting data centers is a primary concern for SDG 7. The reliance on fossil fuels in the energy mix for many data centers exacerbates this issue.

• Energy per Query: Approximately 0.3 to 0.4 watt-hours.
• Daily Inference Energy: Estimated to exceed 340 megawatt-hours (MWh), equivalent to the daily consumption of 30,000 U.S. households.
• Training Energy: The training phase for a model like GPT-3 consumed 1,287 MWh.

Water Consumption and SDG 6 (Clean Water and Sanitation)

Data center cooling is a water-intensive process, placing a direct strain on local water resources and challenging the principles of SDG 6.

• Water per Query Group: An estimated half-liter of water is consumed for every 20 to 50 queries for hardware cooling.
• Projected Industry Use: The AI industry’s global water withdrawal is projected to reach 4.2 to 6.6 billion cubic meters annually by 2027.

Carbon Emissions and SDG 13 (Climate Action)

The platform’s carbon footprint is a direct consequence of its energy consumption, posing a significant challenge to SDG 13.

• CO₂ per Query: Approximately 0.15 grams, depending on the data center’s energy source.
• Monthly Emissions: Estimated to exceed 260,000 kilograms of CO₂, comparable to 260 round-trip flights between New York and London.
• Training Emissions: The training of GPT-3 generated an estimated 550 metric tons of CO₂.

Pathways to Sustainable AI: Aligning with SDG 9 and SDG 12

Achieving sustainable AI requires a multi-faceted approach focused on responsible innovation, production, and consumption, aligning with SDG 9 (Industry, Innovation, and Infrastructure) and SDG 12 (Responsible Consumption and Production).

Key Factors Influencing Environmental Impact

• Prompt Complexity: Complex queries can consume up to 50 times more energy than simple ones, highlighting the need for responsible user practices under SDG 12.
• Model Efficiency: The development of smaller, more efficient models for specific tasks is crucial for reducing resource intensity.
• Data Center Infrastructure: The location, power source, and cooling methods of data centers are critical. A transition to renewable-powered facilities located in cooler climates is essential for meeting SDG 7 and SDG 9.

Industry Response and Recommendations

The AI industry is beginning to address these challenges through various initiatives. To accelerate progress, a concerted effort from developers, providers, and users is required.

1. Invest in Sustainable Infrastructure: Companies must prioritize the construction and use of data centers powered by 100% renewable energy and equipped with water-efficient cooling systems. This directly supports SDG 6, SDG 7, and SDG 9.
2. Promote Model and Hardware Efficiency: Continued investment in energy-efficient chips and the development of less resource-intensive AI models are critical for sustainable scaling, aligning with SDG 12.
3. Enhance Transparency and Reporting: Companies like OpenAI should provide transparent reporting on energy consumption, water usage, and carbon emissions to foster accountability and align with ESG standards.
4. Foster Responsible User Behavior: Educating users on crafting concise and efficient prompts can collectively reduce the platform’s environmental footprint, promoting a culture of responsible consumption (SDG 12).

Which SDGs are addressed or connected to the issues highlighted in the article?

• SDG 6: Clean Water and Sanitation

  The article directly addresses this goal by highlighting the significant water consumption of data centers for cooling AI servers. It discusses the “water usage for AI cooling” and quantifies the amount of water consumed per query and the projected annual water withdrawal by the AI industry.

• SDG 7: Affordable and Clean Energy

  This goal is central to the article, which extensively details the “energy hunger of AI.” It discusses the electricity required for both AI model training and inference, the energy consumption per query, and the importance of the energy source (renewables versus fossil fuels).

• SDG 9: Industry, Innovation and Infrastructure

  The article discusses the rapid growth and innovation in the AI industry (e.g., ChatGPT’s user growth). It simultaneously calls for making this industry and its infrastructure (data centers) sustainable through the use of energy-efficient hardware, better cooling methods, and renewable-powered facilities, which aligns with building resilient and sustainable infrastructure.

• SDG 12: Responsible Consumption and Production

  This goal is addressed through the article’s focus on the environmental footprint of a digital product. It calls for more efficient use of natural resources (energy, water) in the production of AI responses and promotes responsible consumption by users (e.g., using concise prompts) to reduce the environmental impact.

• SDG 13: Climate Action

  The article’s primary concern is the environmental impact of AI, with a major focus on “carbon emissions from AI queries.” It quantifies the CO₂ emissions per query, per month, and from training models, directly linking the technology’s growth to climate change concerns and the need for mitigation strategies like using renewable energy and carbon offsets.

What specific targets under those SDGs can be identified based on the article’s content?

1. SDG 6: Clean Water and Sanitation

   • Target 6.4: By 2030, substantially increase water-use efficiency across all sectors and ensure sustainable withdrawals and supply of freshwater to address water scarcity. The article connects to this by detailing the AI industry’s massive water consumption for cooling (projected to be 4.2 to 6.6 billion cubic meters by 2027) and discussing the need for more efficient cooling methods (advanced air-cooling or closed-loop water systems) to reduce this footprint.

2. SDG 7: Affordable and Clean Energy

   • Target 7.2: By 2030, increase substantially the share of renewable energy in the global energy mix. The article highlights this target by stating that emissions depend on the energy mix and noting that many tech companies have committed to using “renewable energy for data center operations” and aim to run on “100% renewable energy.”
   • Target 7.3: By 2030, double the global rate of improvement in energy efficiency. This is relevant as the article discusses efforts to “improve model efficiency,” invest in “energy-efficient chips,” and reduce the energy needed per query, which are all measures to improve energy efficiency.

3. SDG 9: Industry, Innovation and Infrastructure

   • Target 9.4: By 2030, upgrade infrastructure and retrofit industries to make them sustainable, with increased resource-use efficiency and greater adoption of clean and environmentally sound technologies and industrial processes. The article directly supports this by advocating for “sustainable AI,” which includes deploying “energy-efficient hardware and renewable-powered data centers” and improving cooling methods to make the AI industry’s infrastructure more sustainable.

4. SDG 12: Responsible Consumption and Production

   • Target 12.2: By 2030, achieve the sustainable management and efficient use of natural resources. The article’s entire theme of analyzing and reducing ChatGPT’s environmental footprint—its consumption of electricity and water—is an exercise in promoting the efficient use of these natural resources.
   • Target 12.8: By 2030, ensure that people everywhere have the relevant information and awareness for sustainable development and lifestyles. The article mentions that “user behavior significantly affects environmental costs, making user education part of the solution” and suggests users can help by choosing “concise prompts,” which aligns with raising awareness for sustainable consumption patterns.

5. SDG 13: Climate Action

   • Target 13.3: Improve education, awareness-raising and human and institutional capacity on climate change mitigation, adaptation, impact reduction and early warning. The article itself serves as a tool for this target by educating readers on the carbon footprint of AI. It also mentions industry responses like using “carbon offset programs” and aiming for “carbon-neutral operations” as mitigation efforts.

Are there any indicators mentioned or implied in the article that can be used to measure progress towards the identified targets?

1. For SDG 6 (Clean Water and Sanitation)

   • Indicator for Target 6.4 (Water-use efficiency): The article provides specific metrics that can be used as indicators, such as “water consumption per number of queries” (half a liter of water for every 20 to 50 queries) and “total annual water withdrawal by the AI industry” (projected 4.2 to 6.6 billion cubic meters by 2027). Tracking these figures over time would measure progress.

2. For SDG 7 (Affordable and Clean Energy)

   • Indicator for Target 7.3 (Energy efficiency): The article provides several quantifiable indicators: “electricity consumption per query” (0.3 to 0.4 watt-hours), “daily electricity consumption for inference” (340 MWh), and “energy consumed during model training” (1,287 MWh for GPT-3). A reduction in these values would indicate improved efficiency.
   • Indicator for Target 7.2 (Share of renewable energy): While not providing a current percentage, the article mentions the goal of “100% renewable energy” for data centers. An indicator would be the “percentage of data center operations powered by renewable energy.”

3. For SDG 9 (Industry, Innovation and Infrastructure)

   • Indicator for Target 9.4 (Adoption of clean technologies): The article implies an indicator through its discussion of factors that make AI greener, such as the “adoption of energy-efficient chips” and “advanced air-cooling or closed-loop water systems.” The rate of adoption of these technologies within the industry could serve as a progress indicator.

4. For SDG 12 (Responsible Consumption and Production)

   • Indicator for Target 12.2 (Efficient use of resources): The article highlights that complex prompts can use up to “50 times more energy” than simple ones. This suggests an indicator related to the “average energy consumption per query,” which could be lowered through user education and model optimization.

5. For SDG 13 (Climate Action)

   • Indicator for Target 13.3 (Climate change mitigation): The article provides direct metrics for a carbon footprint indicator: “CO₂ emissions per response” (0.15 grams), “monthly CO₂ emissions” (over 260,000 kilograms), and “CO₂ emissions from training” (550 metric tons for GPT-3). These figures are direct measures of the climate impact and can be tracked to show progress in mitigation.

SDGs, Targets and Indicators Table

Source: carboncredits.com