What Is Ethical AI? Designing AI That Works for People and the Planet

Artificial intelligence (AI) is changing how we work, make decisions and use technology. From healthcare and education to business operations and environmental research, AI can help us solve problems that would otherwise take much more time and effort. But just because AI can do something doesn't always mean it should. 

There's more to using AI responsibly than getting accurate results or following the rules. We also need to think about who benefits from it, who might be left out and what impact it has on the environment. That's the idea behind World Usability Day 2026, which takes place on 12 November under the theme Designing a Sustainable Future with Ethical AI. 

It's also part of the conversation around sustainable AI at COP31, where technology's role in tackling climate change is getting attention. Both raise an important question about the future of AI. How do we make sure the technology we're building is actually good for the people and the world it's supposed to help?

What Is Ethical AI?

Ethical AI means developing and using artificial intelligence in ways that respect people's rights, protect them from harm and take the wider effects of the technology into account. Put simply, it's about making sure AI helps people without putting their safety, privacy, opportunities or the environment at risk. The idea itself isn't complicated. 

The difficult part is making sure it happens in practice. Take an AI recruitment tool, for example. It might be very good at finding candidates who match a job description. But what if it's also more likely to reject people from certain backgrounds because of bias in the data it was trained on? 

The tool might be doing exactly what it was built to do, but that doesn't make the results fair. The same problem comes up when AI makes decisions people can't understand or question, or when a system uses far more energy and computing power than it actually needs. 

UNESCO's Recommendation on the Ethics of Artificial Intelligence, adopted in November 2021, sets out four main values that should guide how AI is developed and used:

• Respect for human rights and human dignity.
• Fair, peaceful and connected societies.
• Diversity and inclusion.
• Protection of the environment and ecosystems.

These values cover things like fairness, privacy, transparency, accountability, human oversight and sustainability. So, while conversations about AI ethics often focus on bias, privacy and regulation, there's an environmental side to it too. And before we can understand how that fits in, we need to look at what ethical design means for the people actually using AI.

What Makes AI Design Ethical?

An AI system doesn't automatically become ethical because the people building it have good intentions. The choices they make about how it works, what information it learns from and how people will use it can all affect the results. 

Ethical AI design means thinking about those effects from the beginning, rather than waiting until something goes wrong and trying to fix it afterwards.

Human rights, fairness and accountability

One of the biggest concerns with AI is that it can repeat unfair patterns that already exist in the real world. AI systems learn from data, and that data isn't always fair or complete. It might contain old biases, leave out certain groups or reflect decisions that weren't fair in the first place. When AI learns from that information, those same problems can show up in its results. 

This becomes especially serious when AI is used to help make decisions about jobs, loans, healthcare or other things that can affect someone's life. Privacy is another part of ethical design. People should know what personal information an AI system collects, why it needs that information and how it's going to be used. 

And if something goes wrong, there needs to be someone who can take responsibility. Human oversight means people should be able to check, question or change important AI decisions when needed, rather than simply accepting whatever the system says.

Accessibility and inclusion

An AI system can be accurate, secure and impressive from a technical point of view, but still be difficult or impossible for some people to use. Imagine a customer service chatbot that doesn't work properly with screen readers, or an AI assistant that struggles to understand different accents, languages or ways of speaking. For the people affected, those aren't small technical problems. 

They could be the difference between being able to use a service and being shut out of it completely. Inclusive AI design means thinking about people with different abilities, backgrounds, languages and levels of experience with technology. It also means remembering that not everyone has a powerful computer, a newer smartphone or reliable internet access. 

This is the thinking behind human-centred AI. Instead of expecting people to change how they do things to suit the technology, we should be designing technology around what people actually need. But people aren't the only ones affected by the choices we make when building AI. 

Even though we usually interact with it through a screen, the technology has a physical impact on the world around us.

What Is Greening AI?

Greening AI means making artificial intelligence more environmentally friendly by reducing the resources it uses throughout its life, from when it's first developed to when it's running and eventually replaced. AI might look like nothing more than software on a screen, but there's a lot of physical infrastructure behind it. 

The computers that power AI use electricity and produce heat, and keeping them cool can require water. Building those computers also takes energy and raw materials, while replacing old equipment creates electronic waste. 

So, when we talk about the environmental impact of AI, we're not just talking about the electricity used to answer a question or generate an image. We also need to think about everything required to make that possible. One way to reduce that impact is to look at whether every task really needs a large, powerful AI model. 

Research published by UNESCO and University College London in July 2025, called Smarter, Smaller, Stronger: Resource-Efficient Generative AI and the Future of Digital Transformation, looked at ways to reduce the energy used by generative AI. The researchers found that changes such as using shorter prompts, making models more efficient and choosing smaller models for suitable tasks could make a big difference. 

UNESCO reported energy savings of up to 90 per cent under the conditions studied. That doesn't mean every AI system can use 90 per cent less energy. The results depend on the model and how it's being used. But it does show that choosing the right AI for a job can make a difference to its environmental impact. 

If a smaller model can do the work just as well, there's little reason to use a much bigger one that needs more energy and computing power. Of course, making AI itself more sustainable is only half of the picture. We can also use AI to help protect the environment.

What Is Greening with AI?

Greening with AI means using artificial intelligence to help solve environmental problems, make better use of natural resources and respond to climate change. The difference between the two ideas is simple. Greening AI is about reducing the environmental impact of the technology itself. 

Greening with AI is about using that technology to help the environment. For example, AI can help researchers study climate data, improve weather forecasts and spot changes in ecosystems. It can also help organisations manage electricity use by predicting when demand will be high and finding ways to use energy more efficiently. 

Other uses include monitoring water supplies, spotting environmental risks and helping communities prepare for natural disasters. These are some of the ideas behind UNESCO's Green Digital Transformation work at COP31, which looks at both making AI more sustainable and using digital technology to support climate action. 

The International Telecommunication Union (ITU) is also developing an AI Climate Institute to help countries use AI responsibly when dealing with climate change, including countries that may not have easy access to the technology or the skills needed to use it. But there's an important catch. 

Just because AI is being used to help the environment doesn't automatically mean it's good for the environment. Imagine an organisation using AI to reduce energy waste. The system might help save electricity, but it also needs electricity to run. To understand whether it's actually helping, the organisation needs to look at both sides. 

It also needs to know whether the system is delivering the results it promised. There's a difference between saying AI could help reduce waste and proving that it has. So, greening with AI isn't just about finding environmental problems that AI could help with. It's about making sure the technology actually makes things better. 

And for organisations thinking about using AI, that starts with some fairly straightforward decisions.

How Can Organisations Make AI More Ethical and Sustainable?

Making AI more ethical and sustainable doesn't have to mean creating a complicated set of rules for every new system. A lot of it comes down to asking the right questions before choosing the technology, making sure it's designed for the people who'll use it and checking that it continues to do what it's supposed to. There are three things organisations should think about.

Start with the problem, not the technology

Before introducing AI, organisations need to be clear about what problem they're trying to solve and whether AI is actually the best way to solve it. It's easy to get caught up in what the technology can do without thinking enough about what people need from it.

• Who's going to use the system?
• What do they need help with?
• Could it make things harder for certain people?
• And what happens if it gets something wrong?

Take a business that wants to use an AI chatbot for customer support. It might be a good way to answer simple questions quickly. But if someone has a complicated or sensitive problem, they should still be able to speak to a person instead of getting stuck in an automated conversation. The same goes for accessibility.

 It's much easier to build a system that works for different people from the beginning than to try to fix those problems later. Sometimes AI will be the right choice. Other times, a simpler solution might work just as well.

Choose AI that fits the task

Once an organisation has decided that AI is the right option, the next question is what kind of AI it actually needs. A large, general-purpose AI model might be useful for complicated tasks that need a lot of knowledge or reasoning. But if the job is relatively simple or specific, a smaller model might be able to do it just as well. 

And that choice can affect more than the results. Different AI models need different amounts of computing power, energy and equipment to run. Some are more expensive to use, while others may need more powerful hardware than the organisation already has. 

So, rather than automatically choosing the most powerful model available, organisations should compare what different options can do and what they need to operate. They should also think about where the AI will run, whether that's on their own equipment or through a cloud provider, and what that means for cost, control and energy use. 

The best choice is the one that can do the job properly without using more resources than necessary.

Check the impact throughout its lifecycle

Even when an AI system has been carefully designed and tested, things can change once people start using it. Users might find problems that weren't obvious during testing. The information the AI relies on might change, or the system might start producing results that aren't as accurate or fair as they were before. 

That's why organisations need to keep checking how their AI works after it's been introduced. The US National Institute of Standards and Technology's AI Risk Management Framework offers guidance on how organisations can identify and manage AI risks throughout a system's life. 

In everyday terms, that means listening to user feedback, checking for unfair results, making sure the system remains accessible and giving people a way to report problems. It also means keeping an eye on the resources the AI uses.

• Is it using more energy than expected?
• Are there more efficient ways to run it?
• Is it still providing enough value to justify what it costs?

An AI system doesn't stop being an organisation's responsibility once it's switched on. It needs to keep working well for the people using it, without creating unnecessary problems along the way.

Final Thoughts: Ethical AI Must Work for People and the Planet

AI can help us do some remarkable things, from making everyday tasks easier to helping researchers tackle problems that affect the whole planet. But being useful isn't the same as being responsible. World Usability Day's focus on ethical AI, together with the environmental conversations around COP31, reminds us that we need to think about more than what AI can achieve. 

We also need to consider who gets to benefit from it, who might be left out and what resources we're using to make it work. As AI becomes part of more of the technology we use every day, those choices will become harder to ignore. And the earlier we start thinking about them, the better chance we have of building systems that genuinely help. 

For organisations trying to make those decisions, keeping up with how AI is being developed, governed and used is part of the job. EM360Tech's coverage of enterprise AI and emerging technology can help make sense of those changes as they happen.