16.4.3. Key Concepts

• Right to Explanation

The 'Right to Explanation' refers to the legal and ethical principle that individuals should be able to obtain clear explanations of how automated decisions that affect them are made. This concept has gained importance in the context of AI, where systems frequently make decisions based on complex algorithms that may not be easily understood by the average user. Essentially, if an AI system makes a decision that has significant impact on someone's life, that person has the right to know how the decision was reached, including the factors that influenced it.

Imagine you're applying for a loan and the bank uses an AI system to determine if you qualify. If the AI denies your application, the 'Right to Explanation' means you should be able to ask the bank to explain how the AI reached its decision. This can include details about the data used, such as your credit score, income, and previous borrowing history.

• AI Ethics Review Boards

AI Ethics Review Boards are interdisciplinary groups tasked with examining the ethical implications of AI systems before they are implemented. These boards typically include experts from various fields—such as technology, law, ethics, and social sciences—to assess whether the AI systems align with ethical standards and societal values. They help ensure that potential risks and biases are identified and mitigated during the development and deployment phases of AI applications.

Think of an AI Ethics Review Board like a group of advisors that a company might hire when developing a new product. Just as a company may consult environmental experts before launching a product to reduce its ecological footprint, tech companies can use these boards to evaluate whether their AI solutions respect human rights, privacy, and fairness.

• Algorithmic Audits

Algorithmic audits are assessments conducted to evaluate the performance, fairness, and impact of AI algorithms. These audits can help identify any biases present in the algorithms and ensure that they operate as intended while adhering to regulatory standards. Algorithmic audits can be carried out internally within an organization or by third-party auditors to provide an objective evaluation. By conducting such audits, organizations can work to improve the robustness and reliability of their AI systems.

Consider a school that regularly reviews its grading system to ensure fairness. Just as schools might audit their grading criteria and processes to ensure no student is unfairly treated, organizations use algorithmic audits to refine their AI systems, ensuring that no particular group is disproportionately disadvantaged by automated decisions.

• Impact Assessments

Impact assessments are systematic evaluations of the potential effects—both positive and negative—that an AI system may have on individuals, society, and the environment. These assessments typically consider various factors, such as ethical implications, potential harms, legal compliance, and tech sustainability. Conducting an impact assessment before deploying an AI system can help organizations minimize negative outcomes and enhance the overall benefits of their applications.

Imagine a city planning to introduce a new high-speed train system. Before starting construction, city planners would conduct an impact assessment to foresee issues related to noise pollution, displacement of communities, and economic benefits. Similarly, organizations developing AI technologies need to assess how their systems will affect users and society at large.

--