The Hidden Environmental and Social Cost of AI

Every human invention comes with a trade-off. The printing press spread knowledge and dismantled power hierarchies. The internet connected us and fractured us. Now we stand at the edge of another shift: artificial intelligence.

At ReCo, we build with AI everyday. It powers our research, shapes our strategy, and helps brands grow smarter. But as usage scales exponentially, we’ve started asking harder questions. How sustainable is AI, really? What are we feeding, and what is it costing us?

The same question was raised at ReCo LIVE 2025, our AI and marketing conference: What are the true environmental, cultural and social costs of using AI?

The question isn’t if AI has impact — it’s how much, and where.

Danling Xiao, ReCo

How many AI queries are made every day?

In December 2024, OpenAI’s CEO shared that ChatGPT was handling over 1 billion queries a day. By July 2025, that number had hit 2.5 billion, and it’s still rising. That’s just one platform.

The question isn’t if AI has impact — it’s how much, and where.

What are the cultural and social costs of AI?

• Homogenisation and echo-chamber effects
  AI is the average of all, and that’s dangerous. We’re risking narrowed worldviews, echo chambers, and a crushing of cultural nuance and diversity. Explained plainly: AI doesn’t discover culture. It upscales what’s already common, drowning out minority voices and localised stories.
  
  
  

• Synthetic empathy: emotional simulations without humanity
  AI can mimic feelings, but that’s not empathy. This synthetic emotional capacity can erode genuine human connection. Over time, we risk substituting lived emotion with polished simulation, sacrificing depth and authenticity in the process.
  
  
  

• Erosion of knowledge systems and power concentration
  AI’s scalability comes at a price: a few tech giants own the platforms, data, and infrastructure. Traditional knowledge systems like indigenous stories, oral histories, artisanal crafts can be sidelined. At the same time, authority over information centralises, distancing communities from knowledge creation and reinforcing existing power imbalances.

The Center for AI Safety also warns of deeper structural risks, including AI’s potential for malicious use, systemic misinformation, workforce displacement and even rogue behaviours from advanced systems. If left unchecked, AI could rewire trust, influence and cultural integrity.

What are the environmental costs of AI?

• Energy: The International Energy Agency warns that data centre electricity use could double by 2026. McKinsey estimates AI could account for 25% of new electricity demand in Europe by 2030.
  
  

• Water: Many AI data centres use evaporative cooling, consuming litres of water per megawatt. In water-stressed regions, that’s a real concern.
  
  

• E-waste: AI hardware depends on rare earth materials, fuelling resource strain and complex waste streams.
  
  

What does a single prompt cost?

Here’s a comparison to put things into context:

• Google search = 0.2g CO2e

• Plain text email = 0.3g CO2e

• Text AI query = 2g CO2e

• Image generation = 35g CO2e

• Video generation = 200g CO2e

According to MIT Technology Review, creating a 5-second AI video is like running a microwave for an hour.

In terms of water usage, each ChatGPT query uses approximately 0.3 milliliters of water. However, research suggests 10 millimetres based on full cycle analysis. (via Medium)

Feels tiny, right? Until you multiply it by 2.5 billion, every single day. And that’s just ChatGPT alone.

Why is AI a power play?

Technology is never neutral. AI is built on human-generated data and reflects the perspectives of its creators.

Behind AI, It’s a race between the US and China, highly equipped with mega investment, compute capacity and data dominance. Both nations have framed AI leadership as a national priority, essential for economic growth, military superiority, and global influence.

Businesses guided by strong ethical values, and those values are more crucial than ever in this movement. Our role is to understand it, empower ourselves and help shape it in a way that aligns with our collective values.

This is why, despite the risks, many inspiring businesses, non-profits, and research groups are actively working to harness AI for good.

What does AI for good look like in practice?

• Radish Events AI Food Waste Tracking Tool by ReCo: Our custom AI-powered image recognition tool enables Radish Events to measure and reduce food waste with 90% accuracy, cutting costs and landfill impact.

• Google Research Flood Forecasting: Using satellite data and machine learning, Google Research helps predict floods up to seven days in advance, giving vulnerable communities critical preparation time.

• Wildbook AI for Wildlife Conservation: Wildbook is an open source software platforms that blend structured wildlife research with artificial intelligence, citizen science, and computer vision.

• Hugging Face AI Energy Score: Hugging Face’s AI Energy Score is an open-source initiative to measure and standardize the energy consumption of AI models, making it easier to build more sustainable and efficient technology.

• Climate Change AI (CCAI): A global organisation of volunteers from academia and industry who are dedicated to catalysing impactful work at the intersection of climate change and machine learning.

• Deon: A command-line tool that adds an ethics checklist to a data science project. It reminds developers to consider ethical issues like data privacy and model bias before they even start building.

How should businesses respond?

1. Use AI consciously: Not every solution needs AI. Ask: can this be done more efficiently with logic, code or creative structure? Only use AI when it adds real value.

2. Offset with impact: The AI tool we built for Radish Events helped them reduce food waste with 90% accuracy, saving time, landfill and operational cost. The impact far outweighs its footprint.

3. Advocate for transparency and ethics: Equip yourself with knowledge and share it with your network. Choose AI models and service providers that are ethical, responsible and sustainable.

4. Support climate action: Ultimately, continue to support climate action. Continue to advocate for awareness, do the amazing work you do, support business and organisations that are at the forefront of climate action.

How can you use AI more consciously?

1. Ask ‘Do I really need AI for this?'

2. Reserve AI for complex problem solving.

3. Choose the lowest-impact tool. Avoid generating content you won’t actually use.

4. Outline ideas, gather data, and refine your questions before asking AI.

5. Group related questions. Avoid ‘one thought, one prompt’ habits.

6. Prefer renewable-powered AI providers, where possible.

7. Preserve spaces for human collaboration and emotional intelligence.

8. Never input confidential or sensitive information. Understand your AI provider’s data retention policies.

9. Read industry reports, books and primary sources. Don’t let AI replace your curiosity and analytical thinking.

10. Always fact check important details before you act or publish.

11. Apply critical thinking for all AI-generated results.

12. Stop asking AI what you should eat tonight!

Lastly, our greatest gift is still human intelligence and collaboration. Inspiration starts with your team, your clients, your customers, and the world around you. Hold that space.

What does using AI responsibly actually mean?

Are we concerned? Yes, 100%. But we’re also a big believer that we must be in it to win it. And that means stay informed, master the knowledge and proactively advocate for our values.

It’s a perennial human struggle to use technology and power wisely and ethically. The choices we make about how we design, regulate, and integrate these technologies into our lives will determine whether they bring us together or push us further apart.