Can the Planet Afford AI?

While listening to a podcast recently, a BBC journalist discussed the AI summit hosted by the UK government and attended by a sea of recognisable faces, including Elon Musk of Tesla fame. When asked what her takeaway from the summit was, the journalist posed a question that she said had received little airtime at the summit - “Can the planet afford AI, and do we really understand the social and environmental costs?”

Which made us think. As technology continues to reshape our world, artificial intelligence (AI) has emerged as a force poised to revolutionise industries, enhance efficiency, and enrich our daily lives. However, the rapid development of AI also raises critical questions about its broader impact. At FuturePlus, we recognise the interplay and relationship between Climate, Environment, Social, Economic, Diversity, and Impact considerations. Sustainability is not a one-size-fits-all concept. It's about understanding the consequences, compromises, and interconnected aspects of a sustainable future. However, as with any new technology, AI has potential risks and unintended consequences, and it is essential to carefully consider the positive and negative impact of AI as we integrate it into our personal and professional lives.

One of the most talked about and pressing concerns regarding AI's environmental impact is its substantial energy consumption. AI systems, particularly deep learning models, require massive computational power, requiring huge amounts of energy and, therefore, a significant carbon footprint. That footprint is currently tricky to measure accurately, but as reported in the MIT Technology Review, there are estimations for OpenAI’s GPT-3 and Meta’s OPT at approx 500 and 75 metric tons of carbon dioxide, respectively, during training.

All that energy creates heat, which requires cooling, and as this study shows, AI can be thirsty, requiring vast quantities of freshwater in cooling systems. Whilst the figures vary, estimations have been made of around 500ml for each AI conversation, spiking some companies' water usage. In a future where water scarcity is becoming an ever bigger issue, innovation will need to continue around AI to help reduce our reliance on energy and water resources.

We also need to build the hardware, and rare earth minerals are key in the technology powering our daily lives. The same technology is critical to addressing the climate crisis, playing a crucial role in various solutions, from renewable energy and electric vehicles (EVs) to artificial intelligence (AI). However, there are long-held concerns about the sourcing of these rare and valuable minerals. Cobalt, a key mineral in the fabrication of batteries and electronic components, is a prime example of these concerns. The extraction of cobalt, often concentrated in certain regions of the Democratic Republic of Congo (accounting for roughly 70 per cent of global production), has for years been marred by reports of unethical mining practices, including the exploitation of child labour and unsafe working conditions. These practices not only have been known to violate human rights but also contribute to environmental degradation, as mining activities often lead to deforestation, water pollution, and soil contamination.

Businesses involved in the production of technology hardware must take responsibility for their environmental and social footprint and implement measures to ensure that their supply chains are ethical and sustainable. This may require investment in cleaner mining technologies, support for initiatives that promote responsible mining practices, and the assurance of workers' rights in mining countries.
AI's social impact also includes concerns for the job market. As Elon said at the AI summit, society could reach a point where “no job is needed” and “you can do a job if you want a job … but the AI will do everything.”

We have been warned that automation driven by AI technologies has the potential to displace millions of jobs across industries, leading to economic and social instability. Workers without access to retraining or upskilling opportunities may bear the brunt of these disruptions. Of course, the opposite may be true, and AI may provide more equitable access to education and training and the opportunity to increase productivity by automating mundane tasks and optimising processes.

Whichever is true, any organisation deploying AI will need to take a comprehensive approach to address its employment and human resource challenges. Initiatives for reskilling and learning programs should be an integral part of the process.

The social implications also extend to data privacy. Chatbots such as such as ChatGPT or Google’s Bard are powered by Large Language Models (LLM). LLMs rely on vast amounts of data, which could be collected from individuals without explicit consent. The potential for misuse, including profiling individuals or discriminatory decisions made by AI systems, raises significant ethical issues, and the legal frameworks are racing to keep up with the innovation. Twinned with this, coding, like any human endeavour, is susceptible to bias. ​​Good AI can help identify and reduce the impact of human biases, but it can also exacerbate the issue by embedding and deploying biases on a large scale. How and what AI is trained on, as well as the transparency and ethics that underpin it, will be vital to avoid discrimination.

The question of whether the planet can afford AI is a critical one. It is evident that businesses integrating AI into their operations will require consideration and deliberation on the trade-offs between the social and environmental costs and benefits as well as the commercial. There is always the possibility of achieving a positive impact in all domains or having to make a strategic choice based on alignment with values, objectives, and the broader business strategy. By doing so, we can all hope that AI becomes a powerful tool for sustainable progress rather than a burden on our people and planet.

Image Credit: Steve Johnson/Unsplash