Artificial intelligence (AI) may be able to achieve all 17 Sustainable Development Goals (SDGs). Although there are some potential downsides, the overall impact of AI on SDGs is overwhelmingly positive with significant benefits for the development of low-carbon sources of energy and efficiency.
The SDGs are a set of 17 interconnected, overlapping global objectives established by the United Nations to address pressing social, economic, and environmental challenges. They aim to eradicate poverty, promote equity, protect the planet, and ensure peace and prosperity for all by 2030. Each of the 17 Sustainable Development Goals has a series of internationally agreed-upon targets as indicated in the 2030 Agenda for Sustainable Development.
How Artificial Intelligence Affects the Sustainable Development Goals
A 2020 study published in the Journal Nature Communications assessed the ability of AI to help us achieve the targets of each of the 17 SDGs. The findings in this research indicate the development of artificial intelligence will have a positive impact on almost 80 percent of SDG targets (134 of 169). However, they also found that AI risks undermining 59 targets (35%). The authors of this study divided the 17 Sustainable Development Goals into three broad interconnected categories: Society, economy, and environment.
Society
The nine social Sustainable Development Goals are SDG 1 (no poverty), SDG 2 (zero hunger), SDG 3 (good health and well-being), SDG 4 (quality education), SDG 5 (gender equality), SDG 6 (clean water and sanitation), SDG 7 (affordable and clean energy), SDG 11 (sustainable cities and communities), and SDG 16 (peace justice and strong institutions).
There are 67 targets or 82 percent of all targets in this group that could potentially benefit from AI-based technologies. However, 31 social targets (38%) may be negatively influenced by AI. The researchers found that artificial intelligence may act as an enabler for all the targets by supporting the provision of food, health, water, and energy services.
Economy
The five Sustainable Development Goals in the economy group are SDG 8 (decent work and economic growth), SDG 9 (industry innovation and infrastructure), SDG 10 (reduced inequality), SDG 12 (responsible production and consumption), and SDG 17 (partnerships for the goals).
AI may have a positive impact on 42 targets or 70 percent of SDG targets in the economy group, while negatively impacting 20 targets or 33 percent of the targets in this group.
Environment
There are three environmental Sustainable Development Goals, they are SDG 13 (climate action), SDG 14 (life below water), and SDG 15 (life on land). AI could help with 25 environment targets or 93 percent of the targets in this category. However, 8 targets may be negatively impacted by artificial intelligence.
AI can analyze large-scale interconnected databases to develop joint actions that preserve the environment. It can advance the understanding of climate change and model possible impacts. AI may also be able to improve the health of entire ecosystems through targets like 14.1 which seeks to prevent and significantly reduce marine pollution and 15.3 which seeks to combat desertification and restore degraded land and soil.
The CHALLENGES and SOLUTIONS ASSOCIATED with APPLYING AI to the SDGs
The authors identified research gaps which they found to be especially pronounced for negatively affected targets. They said that further study is required to ensure that artificial intelligence provides maximum benefits. They also said we should investigate the risks associated with possible failures, preventing malfunctions, and addressing possible hacking vulnerabilities.
The study concludes that to make Ai’s contributions to development truly sustainable we need an inclusive approach that invites the participation of all stakeholders. They also point to concerning issues and some possible solutions.
Unsustainable energy demands: Powering AI with clean energy
We cannot evaluate the value of AI for the Sustainable Development Goals without dealing with the massive energy requirements and large carbon footprints associated with information and communications technologies (ICTs). ICTs could eat up as much as 20 percent of global electricity demand by 2030 which would compromise outcomes in SDG 7 (affordable and clean energy) and SDG 13 (climate action). The study concludes that green growth of ICT technology is essential, including more efficiency, power from clean sources, and embedding human knowledge in the development of AI models. Artificial intelligence can guide the growth of ICT technology, it can also help to address the issue of renewable energy variability by enabling smart grids that match electrical demand to times when the sun is shining, and the wind is blowing.
Lacking technological access: Technology transfer for sustainable development
Those in Low- and middle-income countries who do not have access to AI’s data analytics may be subject to inequalities. This could significantly impact SDGs 8 (decent work and economic growth), 9 (industry, innovation, and infrastructure), and 10 (reduced inequalities).
We also need international agreements to ensure that the wealth that comes from the development of AI does not all go to those who are already advantaged in wealthy countries. Artificial intelligence disproportionately rewards the educated, and this could exacerbate inequality within nations by replacing old jobs with ones requiring more skills. This is why education and technology transfer are necessary to ensure that all countries have access to relevant AI platforms. The authors suggest efforts should go beyond education and technology transfer.
The authors propose a paradigm-changing approach that promotes cooperation. They call for a “global science-driven debate to develop shared principles and legislation among nations and cultures”.
Government inaction: Green policy, legislation and regulation
Policy, legislation, and regulation may be the most impactful ways we can address the problems associated with artificial intelligence. This is the most expedient approach to minimizing adverse consequences, it will also allow us to direct AI toward the areas with the greatest benefit for all three groupings of the SDGs.
We need legislation that incentivizes projects with maximum benefit to ensure that commercially viable projects are not the only ones that garner funding. Understanding the wide array of contexts in which this technology should and should not be deployed is a critical part of crafting effective legislative oversight and regulatory regimes.
Ending opacity: Ethics-focused guidance for applying AI to sustainable development
The authors indicate that we need to carefully assess the societal, ethical, legal, and environmental implications of new AI technologies so that we can craft responsive legislation and certification mechanisms. Legislation must contribute to the transparency and accountability of artificial intelligence, as well as make determinations regarding ethical standards. They further recommend that AI applications that target Sustainable Development Goals should be, “open and explicit about guiding ethical principles and how they align with the existing guidelines.“
Addressing built-in bias: Make AI a viable contributor to the achievement of the SDGs
The authors review the inherent bias in machine learning algorithms. These biases recapitulate harmful stereotypes. This is due to the lack of diversity represented in existing datasets and the word embedding technique used in natural language processing in machine learning. The biased data used to train AI can exacerbate discrimination, inequality, and injustice, particularly towards women and minorities.
Artificial intelligence could also be used to manipulate public opinion for commercial or political gain. Big data can guide deliberately sinister plots designed to willfully control public perception. Sophisticated analytics can be very effective at targeting certain groups with commercial or political propaganda. The power of social media to amplify these messages has been widely documented. AI-driven targeted disinformation known as “big nudging” has proven to be effective at exploiting psychological weaknesses and reinforcing preconceived bias leading to political polarization and a loss of social cohesion.
Big nudging undermines SDG 10 (reduced inequalities), and SDG 5 (gender equality). To address these concerns, we need to understand the ways that artificial intelligence recapitulates biases and thereby discriminates against women and marginalized communities. More study is required to pinpoint the ways that the technology is being used to support nationalism and foster hate toward minorities. We need to be especially mindful of the ways that AI can bias election outcomes. To minimize these devastating impacts, AI must be subjected to ethical scrutiny, transparency, and democratic control.
Combating control and inequality: Diversity and decentralization for equitable sustainable development
The authors advance diversity and decentralization as a bulwark to minimize AI-generated biases and limit the possibilities for control of citizen behavior through artificial intelligence. Diversity and decentralization also have ancillary benefits including supporting innovation and societal resilience.
In practice, the focus on diversity and decentralization means that AI should be adapted to the local values and cultural backgrounds of the places it is being used. As explained by the authors: “We encourage researchers and funders to focus more on designing and developing AI solutions, which respond to localized problems in less wealthy nations and regions”.
How to fix dysfunctional economics: A circular economy is a sustainable economy
There are impediments to a sustainable economy built into our economic system. AI can help us to create a circular economy to address many of these impediments. The authors describe a circular economy as “a multi-currency financial system…which is aligned with societal goals and values”. The study concludes that a circular economy will benefit SDG 7 (affordable and clean energy), SDG 11 (sustainable cities and communities), and SDG 13 (climate action).
DUAL IMPACT of ARTIFICIAL INTELLIGENCE on THE SDGs
The application of AI to the SDGs can have contradictory and even paradoxical impacts. The power of artificial intelligence is in its ability to model the dynamic interplay between technology, people, and governments. It can also help us to navigate the complex interrelationships between the SDGs and between individual targets. However, the interconnectedness of the Sustainable Development Goals and their targets means that success in one area could result in negative effects elsewhere. So, for example, powering AI will require achieving SDG 7 (affordable and clean energy), if we fail to achieve this SDG we will also undermine SDG 13 (climate action). While AI can increase productivity and help us to achieve the full range of SDGs, it can have negative impacts on SDG 1 (no poverty), SDG 2, (zero hunger), and SDG 4 (quality education). AI can also have adverse implications for SDG 5 (gender equality).
There are many ways that the application of AI to the SDGs can have contradictory impacts. Artificial intelligence can help identify sources of inequality, while the conclusions derived from the data it draws upon can be used to manipulate public opinion. Personal healthcare data could radically improve diagnosis and treatments, but it also creates a privacy vulnerability. Although AI is increasingly applied to improve biodiversity monitoring and conservation, AI-related ecosystem data may contribute to the over-exploitation of resources.
The complex interconnections between Sustainable Development Goals demand that we exercise caution when using AI. We need further study, particularly in areas where there is little data to draw on. So, for example, artificial intelligence offers tremendous benefits for the achievement of target 14.5 on conserving coastal and marine areas, however, the effects on equity and fairness are unknown. Similarly, SDG 16 (strong and connected institutions) is critical to regulating the future of AI, but our understanding of the influence of AI on institutions is limited. We also need to develop a better understanding of how to best use data mining and machine learning as predictive tools.
CONCLUSION
Lured by a wealth of benefits, we will become ever more dependent on sustainable-development-friendly AI. Although artificial intelligence has an overwhelmingly positive impact on the SDGs, we cannot afford to ignore the negative effects. Addressing these negative impacts is essential to maximizing the benefits of AI. There are things that AI does exceptionally well like fraud detection and assessing the possible effects of legislation. But we need to be wary of the ways in which AI can be counterproductive.
Taken as a whole, AI can be a powerful tool for advancing sustainability. We have only begun to scratch the surface of what AI can do for sustainable development. AI can be paired with existing technologies like satellite imaging, it can also be paired with new and emerging technologies that will exponentially increase the benefits of the Sustainable Development Goals.
It is important to point out that this is only one study, and more research is required to see if these results can be replicated. Given the complex interconnected nature of the SDGs, future research may benefit from factor analysis as well as novel methodologies.
In the absence of understanding and oversight, AI will be used to manipulate public opinion in a way that serves corporate or political agendas rather than provide benefits to people and the planet. However, with the proper study, we can craft legislation and regulations, that guide the creation of an ethical, responsible, and trustworthy artificial intelligence that could improve our lives in ways we have not yet imagined.