Tech for sustainability, or the sustainability of tech?
Tech and sustainability – two words we have come to commonly hear used together. They both play an important role in the development of one another. More recently, we have seen tech being used for good and for the purpose of enhancing sustainability, but is the technology we use sustainable enough? Rather than focusing solely on the positive implications it has, we should be looking to how sustainable it truly is. 
It is important to start right at the beginning. A mobile phone’s biggest environmental impact comes from the manufacturing process, which produces the equivalent of at least 50kg of CO2 per device.  Most people would look to the technology leaders to stop producing so many devices, but they are only responding to the demand from their consumers. We live in a capitalist society where everyone wants the best of the best. As soon as new devices are released, some people are guaranteed to ditch their old models regardless of how long they have had them. With how quickly new designs are released, this can be as short as a year. Consumers can take it upon themselves to extend the lifespan of their devices – why fix something if it is not broken? Rather than upgrading after two or three years, keep it for five. This is the most effective way to reduce the harmful effects the large volumes of CO2 have on the planet. Simple DIY repairs, replacing parts when they have reached the end of their lifespan, and seeking support from tech experts are the best ways to keep devices going for as long as possible. 
Artificial Intelligence (AI) is ever-growing. Alike to technology, not only do we need to continue using it for good, but ensure it is sustainable and does not have negative impacts on our beloved planet. Starting at the beginning of AI’s lifespan also leads us to the supply chain it is produced in. If those producing AI continually create applications that are only compatible with the latest operating systems and devices, it will only add to the economic waste the technology industry is continuously fabricating. In turn, this will create issues for those who refuse to conform to the need to have the latest device, or simply cannot afford to. Equality is key and not accommodating for those with older technology will undoubtedly put them at a disadvantage.
For example, AI for Good have created a chatbot to prevent domestic abuse and violence in relationships. Users can read stories, take quizzes and look at resources to understand and identify if their relationship is violent or abusive.  For vulnerable people in disadvantaged countries where vital education on gender-based violence is not provided, the use of an AI chatbot is invaluable. By ensuring this is available on older devices and software, everyone has the access they need, and an economic bias has not been perpetuated.
In today’s society, it is difficult not to be glued to our technology. Working from home has made it more difficult to detach from our phones, laptops and computers than ever before. Checking social media, responding to emails and keeping in touch with family and friends are things we cannot do without, but if users knew how much of a negative environmental impact simply checking their devices has, we may see the much needed change our world desires. On average, we pick up our phones 58 times day.  Each time we do, the amount of greenhouse gases released builds and builds. Unsurprisingly, with over 4.5 billion people online, using our phones and the systems that allow them to run contributes to 3.7% of all greenhouse gas emissions – which is around the same as the airline industry.  As a result, by 2040, communication technology will account for 14% of the global environmental footprint. 
AI is particularly data and energy greedy. Just like electronic waste, many individuals do not know what happens to their data after three years. In the past, some service providers destroyed all of their hard drives for security reasons when switching to the cloud. However, this process poses a number of important questions – Is it deleted safely in accordance with GDPR? Are the servers destroyed or recycled when the data retention license expires? And do these companies have an environmentally friendly server upcycling or data sanitation process? We need to begin getting answers to these questions from tech leaders or we will not see the progress needed.
Building complex algorithms with more data is not the answer to being more sustainable. Instead, creating something that is trained in a small set of data and can continue to learn over time is the more intelligent way to go. Imagine if all companies were training AI to do unnecessary tasks – the only impact this would have is creating more servers and consuming more energy, which will only end badly. Good progress has been made through this approach and, over the next decade, we should stick to only producing AI that is going to make a real difference.
In the technology sector, particularly within AI, duplication needs to be avoided. Producing the same thing over and over again is unnecessary. Within technology, producing a new device with a clear difference to its competitors can be justified, but there is a lot of symmetry in the information circulating in the humanitarian and public goods sectors of AI. A lot of money will have been spent customising the AI to ensure it is ethical and sustainable, so why have we not developed a habit of openly sharing and learning from the best in the practice? Building the same complex solution from scratch time and time again is not the answer. In order to find a global scale solution to some of the toughest societal challenges, such as climate change, equality and coronavirus, we need to act quickly and collaborate.
Despite issues with supply chains, data storage and energy usage, and duplication, change is happening. In September 2019, “Google pledged to achieve zero emissions by 2030, while also striking 18 wind and solar energy deals”, which they called the ‘biggest corporate purchase of renewable energy in history’.  Following in Google’s footsteps, in July 2020, Apple also committed to being carbon neutral across the entire business by 2030. “If Apple’s target is met, it would be the equivalent of taking three million cars off the road each year.” 
Microsoft’s data centres have been 100% renewable since 2014 and they have invested $1 billion in the development of carbon neutral technologies. Not only do they intend to be net zero by 2030, but to be carbon negative and, impressively, to remove enough carbon by 2050 to balance all the emissions they have released since they were founded in 1975. 
Microsoft are leading by example and their internal carbon fee could be a point of inspiration for many companies. Tortoise Media revealed “since 2012, the company has operated an internal charge, now set at $15 per tonne, to incentivise its business units to reduce their carbon emissions. Proceeds pay for sustainability improvements and, from this year, will also be applied to emissions from their suppliers.”  They are driving the technology world to a more sustainable future through preventative measures, as well as fixing the negative impacts they have had in the past.
There has also been progress in the techniques used to reduce the size of AI models before they are in widespread use. ‘Model compression’, a variant of the ‘pruning’ technique, sees neural networks undergo training and have their weakest neuron connections removed. While the performance of the model will be the exact same, it will no longer be as large.  This technique may spark a change in how energy and data greedy AI is proving to be at the moment.
Although there is an assumption that being sustainable and using renewables comes at a higher price, in the long run we are all going to benefit. We can all start by making small changes to our daily technology habits, but it is the technology leaders who should be taking a stand as they have the resources and funding to have the biggest impact. A lot of key players in the tech industry have already stepped up the challenge, namely in terms of reducing waste generation and emissions, but it does not stop there. Actions speak louder than words and we must hold companies accountable for their actions and to their short and long-term promises. Focusing on both the sustainability of technology, and the use of technology for sustainability is the answer to the worst problems our society faces.