AI’s Role in Sustainability

This month, I attended Sustainability Live Net Zero 2024 where, unsurprisingly, AI was a big topic of discussion. Speakers delved into the topic of how AI will help - and possibly hinder - sustainability efforts within businesses. It was eye-opening to realise that whilst we’re all excited about the development of new AI tools and how they can be used in our day-to-day lives, not many people are talking about the environmental impact of these tools.

The Cost vs Benefits Debate

The biggest debate around the topic of AI and sustainability is whether or not the benefits will outweigh the cost. As it stands, the current AI tools use huge amounts of energy (a single chatGPT request is estimated to use 0.3kWh of energy, whereas a Google search uses 0.0003kWh), and require lots of large data centres to power them. As AI gets better, the amount of energy needed to power it is increasing. Data sets being used to train the models are expanding, and therefore consuming more energy. It is a possibility that AI will be able to come up with solutions to these problems, but there is no way of knowing what the quality or applicability of these solutions will be. It’s an interesting and complex topic of discussion that everyone needs to keep in mind when we think about integrating AI into our everyday lives.

As AI becomes more popular, we collectively need to build our use of AI around the sustainability principles that are already in place, rather than using AI with no sustainability regulations in mind, and then having to clean up the mess later. The question remains whether or not companies will be able to successfully leverage AI to help contribute to sustainability initiatives.

Optimisation & Efficiency

AI could be leveraged for increased efficiency in two big ways that could influence sustainability within businesses. First, for optimising workflow. We are already seeing how AI is able to do some tasks which are currently done by humans, and whilst this can be worrying in some ways, it also has the potential to increase human capacity by freeing up the time of engineers, designers and others to work on sustainability initiatives and continue with human innovation that currently can’t be replicated by machines. Secondly, by helping with smart and efficient design of products. We may be looking at a future where AI can help drive progress and innovation much faster than humans could. For the world of digital products, this means helping to create products that are more efficient, easier to use, and built in a way which is well-organised and therefore has less of an impact on carbon emissions.

Education & Better Resources

For businesses who are trying to introduce more sustainability initiatives, education is key. It can be hard to educate employees on sustainability if it’s unknown what resources are available. Collating information for training programmes, resources and workshops are all tasks that could be done by AI in the future. 

Ethics & Responsibility

Ethics and responsibility is a hot topic within the AI space, as we don’t have the information to know exactly how this relatively new technology will progress. Ethics and responsibility also go hand in hand with sustainability and environmental policies, so when these two come together, there are a whole host of considerations around how and even if we can use AI responsibly for sustainability. 

For machine learning, there needs to be something in place that regulates where the data sets come from, and ensures that there is a wide variety of good quality data. There are already issues with greenwashing, and this could be made much more complex with the potential of companies to be able to train AI to influence the outcomes of models for their own benefit.

Developing good AI tools is expensive and time-consuming. Consideration should be given to how we can make sure that it isn’t just the big companies that have access to these tools. In the context of sustainability, and AI being used for good, we have to ask whether or not larger companies will be willing to open source some of these for the greater cause of saving our environment.

Data waste and management

Data waste is one of the biggest contributors to the large carbon footprint of AI, as well as a general contributor to many organisations overall carbon footprints. It was found that around 68% of corporate data is never used, so we have to ask ourselves if this will be made worse with all of the data also used for AI. Machine learning requires large data sets to train the models, which takes up huge amounts of energy from data centres. On the upside, it seems feasible that AI will be able to come up with better solutions for managing data than what we currently have.

Ultimately, there are already struggles with being able to support the data that we are currently using. The exponential growth in data is causing a strain on electricity supplies in countries around the world. Governments are looking to put restraints on this digital infrastructure growth, for example, in Ireland, EirGrid put a stop to plans for up to 30 new data centres to be built until the country can find a way to meet their electricity needs with renewables. With the introduction of bigger and more complex AI models, the question remains, will we be able to support this growth whilst keeping renewable energy solutions in place?

What is the value for ROI?‍

Arguably we are currently in the ‘gold rush’ era of AI. This means that everyone is currently doing their own experimentation with it, and figuring out the ways that it can benefit their organisations. Hopefully, some great environmental solutions will come out of this experimentation, but unfortunately it means that there will be short-term consequences for potential long-term gain. The question is - will this long-term gain be able to mitigate the short-term effects?

For most companies, return on investment is usually measured in money and in growth, but regarding AI, does the value for ROI need to change? As the new generation enters the workforce, surveys are showing that a huge percentage of them are caring more about the ethics of a company that they are considering working for, than the pay that they are offering. So, perhaps the mindset around the ROI on AI needs to be shifting from purely monetary to how being responsible with the use of AI may attract new talent to the business, and encourage more innovation from people who are motivated by a genuine passion for finding solutions to combat the climate crisis.

Conclusion

Ultimately, the question that needs to be answered is: can AI be used for sustainability efficiently enough to mitigate the effects of using the AI in the first place? And as of right now, the answer is that nobody knows. All we can do at the moment is continue to talk about the environmental impact of AI and try to encourage people to use it as responsibly as possible. This isn't about stopping or hindering growth and progression, but about asking ourselves how quickly this growth really needs to happen and how we can progress responsibly and more sustainably.

‍