ICTworks

There is little doubt that artificial intelligence technologies will be transformational. Breathtaking advances will be made, extraordinary wealth will be created, and many of our social and institutional structures will be transformed. However, we must ask: whose lives will be improved (or harmed) by these technologies?

A key assertion of this paper, “Artificial Intelligence and Human Development” is that, if we continue blindly forward, we should expect to see increased inequality alongside economic disruption, social unrest, and in some cases, political instability, with the technologically disadvantaged and underrepresented faring the worst.

The Artificial Intelligence Divide

This prediction stems from the interweaving of two elements: the nature of AI applications, and projections of the impacts of AI applications in the current global context. What is worrisome is the dynamic of how our current set of institutions and cultures shapes the evolution of technologies, and how, in turn, these technologies shape these institutions and cultures.

One important element of this context is that it is characterized by what can be called an “AI divide” – that is, a gap between those who have the ability to design and deploy AI applications, and those who do not. Furthermore, both the digital (e.g., infrastructure) and analogue (e.g., regulations) foundations required for an ethical and equitable application of AI technologies in many countries in the Global South are largely absent, and salient power asymmetries persist.

This situation makes it all the more important to address the challenges posed by AI so that we can avoid or mitigate these adverse outcomes while enabling developing countries to take full advantage of AI’s positive potential. The promise of AI is too alluring, and its potential too great to avoid an AI future. The question is whether or not we will be ready.

Artificial Intelligence Benefits for LMICs

Health care: AI can play a crucial role in augmenting health care capacity by filling gaps in human expertise, improving productivity, and enhancing disease surveillance. For example, an NGO in Brazil has partnered with an AI start-up to develop a system to predict upcoming incidences of disease.

Government services and information: Groups around the world are exploring ways to use AI to help countries improve their e-government efforts by automating complex assessments that take account of a range of technical, organizational, and social factors. For example, a machine learning system has been developed to help predict mass grave locations of Mexican drug cartel victims.

Agriculture: AI is being employed to address the various threats that can compromise a successful harvest. For example, AI systems are being used to support water management in Palestine and drought monitoring in Uganda.

Education: AI can move educational offerings beyond an industrial, one-size- fits-all delivery model toward quality personalized learning opportunities at scale. For example, efforts in India are employing AI to develop intelligent tutoring systems.

Economy and business: AI offers the potential for higher productivity and offers a means of growth in the form of new business development, innovation, and optimization of economic building blocks. For example, several companies are working to extend access to standard financial services to the hundreds of millions of Africans who either do not use them or do not currently have access.

Artificial Intelligence Risks for LMICs

Fairness, bias and accountability: AI systems have the potential to reflect and exacerbate societal bias and produce results that can disadvantage individuals and groups, especially those already marginalized. For example, a computer program used in the U.S. to assess the risk of re-offense by individuals in the criminal justice system flagged black defendants as high risk nearly twice as often as white defendants.

Surveillance and loss of privacy: AI algorithms supercharge surveillance and threaten privacy. For example, AI-powered facial recognition software gives closed-circuit TV systems the capacity to track individuals as they move through the urban landscape. This is concerning both socially and politically, as privacy is key to other fundamental rights such as freedom of expression and association.

Job and tax revenue loss through automation: With the growing use of machine learning and AI systems in nearly all sectors of the economy, widespread automation will extend beyond manufacturing to impact higher- skilled knowledge-based roles. Many of these jobs can be partly or entirely automated, reducing the need for human workers. However, a counter- argument has also been made: AI may shift the nature and scope of work and jobs, for instance through robots complementing human labour, and an increased focus on higher-skilled and higher-paid tasks.

Undermining democracy and political self-determination: In a world increasingly connected and reliant on the free flow of information, misinformation is a genuine and growing threat to stability and democracy. For example, by piggybacking on highly personal data collected on social media campaigns, AI applications facilitate more efficient propaganda and behavioural manipulation campaigns. The 2016 U.S. presidential election has become a notorious example of the role of targeted misinformation over Facebook.

A lightly edited excerpt from Artificial Intelligence and Human Development by Matthew L. Smith, International Development Research Centre, and Sujaya Neupane, McGovern Institute for Brain Research, MIT.