top of page

As artificial intelligence becomes more prevalent, assessing its real-world impact becomes more essential. The people designing data-powered artificial intelligence systems are increasingly aware of their power to have transformative, long-term impacts. The Canadian government, for example, has introduced an “algorithmic impact assessment” tool, to help determine the potential real-world impacts of automated decision-making systems used for tasks like visa application processing. Some have questioned the transparency of governments (and businesses) when it comes to the use of AI to deliver services; in 2018, officials in New Zealand announced a “stocktake” of how government agencies were using algorithms to analyse data, amid concerns about potential racial profiling in automated visa application processing. Impact assessment is integral to responsible data governance. There are several methods for quantifying the impacts of technologies and policies, for both the private and public sectors. There are also qualitative methods for better understanding how technology and data affect different populations differently - these can generally be relied upon to help ensure the effective identification and participation of different stakeholders, and to evaluate the effectiveness of systems in terms of supporting an organization’s stated values. These methods can also be used to evaluate the appropriateness of policy and regulatory responses (in the form of standards and guidelines), to better understand social and environmental ramifications of data practices, and to divine the values and norms that can better promote positive aims. The speed at which emerging technologies and new datasets develop means that evaluating impact - whether positive or negative - can be fraught with difficulty. Addressing basic questions such as what constitutes “good enough” evidence, and what mechanisms are available to ensure the sharing of critical evidence, is essential for the legitimacy of impact assessment efforts. Increasingly complex and dynamic social systems, incomplete and often siloed datasets, and methodological constraints can only further complicate matters. Regardless, impact evaluations and assessments are important tools that can help build public confidence in the design and deployment of data systems. This will only become more important, in light of the increasingly prominent role artificial intelligence is playing in maintaining global stability, implementing adequate cybersecurity, managing global power dynamics, and maintaining international security.

The Impact and Implications of Data


Data Era