The Environmental Impact Of GenAI Doesn't Have To Be Disastrous

Generative Artificial Intelligence (GenAI) has become an integral part of everyday life. Software like ChatGPT and Gemini enables users to create content, such as images, text, music, and videos, thatmimics human cognitive functions. These technologies have reshaped our relationship with technology, redefinedthe role of humans in creative processes, and raised serious questions about environmental sustainability.

Arecent study analyzing the role of Artificial Intelligence in Africa's outsourcing and tech-enabled services industry revealed that 40 percent of the jobs in this industry are at high risk of automation.

The findings are particularly concerning for specific demographics. "Junior roles, which are most common in the sector, are highly susceptible to automation, with over half of the tasks automatable. Women, who disproportionately occupy these roles, face a 10 percent higher automation risk than men," the study reports. It further cautions that as AI takes on more complex tasks, entry-level workers must develop beyond basic digital literacy, requiring foundational knowledge of AI and technical proficiency to work effectively alongside these systems.

Environmental cost

GenAI's environmental footprint extends across multiple natural resources. Building large data centers that support these applications requires substantial physical space. Cooling the systems demands significant water resources. Deforestation occurs during the development of data centers, and hardware production relies on rare-earth materials.

Research has found that the carbon footprint required to train one large AI model is equivalent to five cars over their lifetimes. Furthermore, the mining processes for rare-earth materials needed in AI hardware cause significant environmental degradation.

Oghosa Erhahon, a Policy Officer at Long Duration Energy Storage (LDES) Council, emphasizes this reality. "When it comes to Artificial Intelligence, to have connectivity, you need infrastructure, and infrastructure puts pressure on the environment, supply chains, and even on the health of people living around these facilities," she tells OkayAfrica.

Preparing Africa

African tech entrepreneurs have actively participated in the AI revolution, building AI-powered solutions and integrating AI features into existing operations. A significant milestone was reached in March whenCassavaTechnologies partnered with NVIDIA to establish the continent's first AI factory, with plans for supercomputer deployments in Nigeria, South Africa, Kenya, Morocco, and Egypt.

However, this progress comes with caution. Even in technologically advanced countries, AI infrastructure has caused environmental damage, affecting natural landscapes, displacing communities, and creating water and noise pollution. For Africa, sustainable implementation is crucial. The negative impact could be disastrous if data centers on the continent are not designed with a strong emphasis on energy-efficient hardware and reduced reliance on rare-earth materials.

"The biggest issue we would have is connectivity because we don't have a lot of localized connection infrastructure," Erhahon says. "Even if there was supposed to be a digital center in Nigeria, they would have to own a decentralized energy resource system as it couldn't rely on the existing power grid."

For users who increasingly rely on AI for complex and simple tasks, more thoughtful usage is crucial. Optimizing interactions through better prompts and limiting use to vital applications can reduce environmental impact.

"Those kinds of frivolous searches, I don't think that's what the infrastructure for artificial intelligence was made for," Erhahon observes, highlighting how AI-generated internet trends can needlessly increase resource consumption. "Because there's a rampant push to use AI and excuse ourselves from thinking critically, we're wasting a lot of energy in how we use it. And I think that will harm the planet down the road."

Ultimately, ensuring that AI has a minimal effect on the environment and the planet's future is both an infrastructural and personal issue. "It's something that we all have to think about. We need to ensure that the future is also dependent on low-carbon fuels because if we want to get into this futuristic era, we also need to reduce our carbon footprints in the process," Erhahon says.

This reduction in carbon footprints involves rethinking supply chain management, mining for rare earth elements, transporting machinery to data centers, and other related activities.

A push for renewable data centers, increased regulations around environmental impacts, thorough environmental impact assessments before developments, Green AI development, and a clear understanding of what it costs to generate a single photo or a prompt will be crucial in safeguarding the planet's future.