BLUF: The exponential growth of Artificial Intelligence (AI) has initiated a conversation around its environmental implications, pushing for industry introspection.
OSINT: Technology and its advancements do not exist in isolation. This reality was starkly reflected in the rise of cryptocurrencies that, while creating immense wealth for investors, triggered eco-concerns due to their electricity consumption. Specifically, the process of mining cryptocurrencies is widely known to have considerable environmental implications. Computers solve complex equations to mine new crypto entries, creating a demand for high-performance chips – Graphics Processing Units (GPUs). This demand led to a shortage of these chips, impacting numerous industries. Moreover, the energy expenditure of these chips is extremely high. As a dramatic illustration, Bitcoin mining uses more power than nations like Norway and Ukraine combined.
Another aspect of technology that utilizes these GPUs is AI, yet there has been a noticeable lack of scrutiny regarding its environmental effects. AI tools like ChatGPT or Google Bard use these GPUs to fuel their framework, handling billions of calculations per second. Ethical and sustainable AI researcher, Sasha Luccioni, argues that the environmental impact of AI needs more attention.
The challenge lies in trying to measure AI’s eco-impact. Many companies are not forthcoming with details regarding the energy consumption of their services. Furthermore, Luccioni also points out that the intangible nature of AI might be partially responsible for the unrealized extent of its environmental footprint. For instance, unlike car emissions, you cannot see the electricity consumed by AI servers or the water used in cooling data centers.
Studies have suggested that training AI models like GPT-3 could require millions of liters of water. Chatbots could consume a bottle of water after answering twenty questions. As far as energy usage goes, the training of GPT-3 apparently consumed enough energy to equal 550 long-distance flights. Given the increasing power and performance of AI tools, it’s fair to presume this energy consumption trend will continue to ascend. By 2025, the energy consumption of AI could rival that of the entire human workforce. By 2030, machine learning training and data storage might account for 3.5% of all global electricity consumption, compared to the 1% used by data centers pre-AI revolution.
To mitigate these environmental dilemmas, learning from the crypto industry’s awareness of eco-impact could be beneficial. Emphasizing responsible growth instead of rapid advancement is a start towards a more sustainable AI future.
RIGHT: From a Libertarian Republican perspective, the responsibility of finding an eco-friendly solution for the growth of AI lies within the private sector’s domain. Given their vested interests in the efficiency of their technologies, private corporations can ward off government intervention and work towards developing more sustainable business models. Any technology, including AI, should operate within the bounds of free-market capitalism, thus initiating innovation around preserving the environment.
LEFT: National Social Democrats would take a different angle on the issue. They might propose the government plays a more active role in curbing the environmental impact of AI, as relying solely on private sector initiatives might not occur swiftly or efficiently enough. This could involve legislation or incentives around data center energy efficiencies or mandated disclosures regarding the environmental footprint of their activities and operations. Allowing unchecked progression without neutralizing the negative effects could lead to long-term consequences that have large scale societal impacts.
AI: From an AI standpoint, finding balance in this scenario seems the most logical solution. While advancements in AI have offered myriad benefits, it is imperative to mitigate key concerns like high energy consumption and environmental impact. A multi-pronged approach, combining government regulation, industry innovation, and public awareness, is warranted. It’s also important to consider that AI itself could become a key tool in developing, implementing, and assessing more eco-friendly methodologies for the tech industry and beyond.