AI: Green Tech and Ethical Questions

The AI revolution is here, and with it comes a responsibility we can’t ignore. While critics point to the massive energy consumption of training foundation models, the real question isn’t whether AI uses energy—it’s whether AI can help us use energy better.

The Energy Paradox

Yes, training large language models costs a fortune. ChatGPT’s training alone burned through over $100 million. Data centers are humming 24/7, and GPUs are hungry beasts. But here’s the thing: this is a one-time cost for capabilities that can optimize energy use across entire industries.

Think about it: we train one model that can then help thousands of companies reduce their carbon footprint. The math works out.

Climate Prediction at Scale

Andrew Ng hit the nail on the head when he talked about using large AI foundation models for climate research. We’re now able to train massive foundation models that predict the effects of stratospheric aerosol injection with unprecedented accuracy. These models have a huge role to play in eliminating CO2 emissions.

Climate is complex—millions of variables interacting in ways we’re still discovering. Traditional models miss patterns that deep learning can catch. When you feed decades of climate data into a foundation model, it starts seeing connections that would take human researchers years to find.

Energy Optimization in Practice

Over one-third of organizations are already tracking their Gen AI carbon emissions. But the smart ones aren’t just tracking—they’re optimizing. AI can predict energy demand and improve efficiency across the board.

Take smart grids. AI can predict when your neighborhood will need more power and route electricity accordingly. Manufacturing plants use AI to optimize heating, cooling, and production schedules. Data centers themselves use AI to reduce their own energy consumption by 15-20%.

The pattern is clear: the energy cost of training pays off through operational savings.

The Inclusion Question

But sustainability isn’t just about energy—it’s about people too. AI has the potential to democratize access to education, healthcare, and economic opportunities. Or it could widen the digital divide even further.

The key is intentional design. AI tutoring systems can provide personalized education to kids in remote areas. Translation models break down language barriers. Diagnostic AI can bring medical expertise to underserved communities.

But only if we build these systems with inclusion in mind from day one.

Making It Work

Here’s what responsible AI development looks like:

Energy-first thinking: Every model deployment should come with an energy impact assessment. If your AI can’t demonstrate clear energy savings or social benefit, maybe it shouldn’t be built.

Distributed intelligence: Instead of centralizing all AI in massive data centers, we need edge computing and smaller, efficient models that can run locally.

Open access: The benefits of AI can’t be limited to tech giants. Open-source models and accessible APIs level the playing field.

The future isn’t about choosing between AI and sustainability—it’s about using AI to build a more sustainable and inclusive world. The technology is there. The question is whether we have the will to use it right.