The intersection of artificial intelligence and governance has become a focal point of debate, particularly concerning potential deployment within the United States government. Concerns have been raised by experts regarding the lack of transparency, potential for bias, and the overall implications of entrusting critical governmental functions to AI systems without proper oversight and validation.

One of the primary worries stems from the lack of clarity surrounding how AI systems are being developed, tested, and implemented within the government. Experts emphasize the importance of designing AI tools with specific purposes in mind and rigorously testing them to ensure accuracy and fairness. However, there are concerns that these crucial steps may be overlooked in the rush to adopt AI technologies.

Shobita Parthasarathy, a professor of public policy at the University of Michigan, highlights the dangers of blindly trusting AI systems without understanding their underlying algorithms, training data, and decision-making processes. She warns that without such knowledge, there is no basis to believe in the trustworthiness of AI-driven decisions.

Cary Coglianese, a professor of law and political science at the University of Pennsylvania, expresses skepticism about using AI to make decisions about employment, citing the potential for mistakes, bias, and other problems. He argues that such applications of AI could have severe consequences for individuals and communities.

These concerns are not limited to theoretical scenarios. Examples from around the world demonstrate the potential pitfalls of poorly executed AI systems in government. In the Netherlands and the United Kingdom, flawed AI systems have led to the wrongful denial of welfare benefits to residents in need. Similarly, in the United States, the state of Michigan faced a crisis when an AI system used to detect fraud in its unemployment system incorrectly identified thousands of cases, resulting in unjust penalties and accusations against innocent individuals.

Hilke Schellmann, a professor of journalism at New York University and an expert on artificial intelligence, emphasizes the need for a fair, just, and equitable use of AI in government. She cautions that the rush to adopt AI technologies could result in undetected harms, particularly for low-income and marginalized communities who often interact with government through social services.

One of the challenges of using AI to replace workers in the federal government is the diversity of jobs and the specialized skills required for each. An AI program would need to be incredibly complex and highly trained to perform even a mediocre job at replacing a human worker. Coglianese notes that the tasks performed by government employees are often highly specialized and specific, making it difficult to simply replace them with AI.

Parthasarathy argues that AI does not possess the real expertise and nuanced understanding of issues that human workers have. She explains that AI is merely a tool for finding patterns based on historical data and is likely to have limited utility and even reinforce historical biases. In fact, AI doesn’t ‘understand’ anything.

Despite these concerns, experts acknowledge that AI can be a valuable tool when developed and deployed responsibly. AI can automate repetitive tasks, freeing up workers to focus on more important matters, and assist workers in solving complex problems. However, it is essential to proceed cautiously and ensure that AI is implemented in a way that benefits society as a whole.

Starlink, SpaceX’s Low-Earth Orbit (LEO) satellite internet service, is now operational in Niger after the country’s military government granted it a five-year license to provide high-speed connectivity. This makes Niger the latest African country to approve Starlink, following its recent launch in Liberia in January. Niger has one of the lowest internet penetration rates in Africa, with only 32% of its population having online access. The country’s digital infrastructure remains underdeveloped, covering just 30% of its territory, leaving vast rural areas disconnected. This limited access is partly due to low investment in the sector and ongoing attacks on communication infrastructure by armed groups, according to the electronic communications regulator ARCEP. By leveraging satellite technology, Starlink aims to bridge this gap and provide fast, reliable internet across the country.

Starlink’s expansion in Africa is part of a broader trend to improve connectivity on a continent where only 40% of the 1.3 billion people have internet access, the lowest rate globally. However, the service has encountered obstacles in some nations. For example, despite its presence in neighbouring countries, Starlink remains absent in South Africa, Africa’s largest and most industrialized economy. In August 2023, South Africa banned the import of Starlink kits, leaving the country cut off from SpaceX’s satellite internet. Talks to introduce the service in the country have repeatedly stalled, even as demand for alternative internet solutions continues to grow. Elon Musk recently claimed that Starlink is not allowed to operate in South Africa because he is "not Black." Starlink, which serves over 2.6 million customers worldwide, provides broadband internet through a network of approximately 5,500 satellites that SpaceX started deploying in 2019.