AI and society

A societal claim should identify the people affected and the consequences being discussed. Ask whether the claim concerns a person, a community, an organization, or a wider public; whether it describes a benefit, harm, or trade-off; and which stage of designing, deploying, or using AI it concerns. NIST describes trustworthiness characteristics as socio-technical and says they must be balanced according to the system’s context of use. These characteristics can help frame questions, but they do not by themselves establish what happened in a particular case. For that, look for evidence about the system and its context of use.

Try this distinction: a stated value tells you what matters; evidence helps establish whether a particular outcome occurred.

• Artificial Intelligence for a Better Future

  An open access book exploring AI ethics through social and organizational questions.

Claims about AI often combine an observation with an interpretation: for example, a system produced a particular output, so a service must have improved. To assess the conclusion, identify the outcome being measured, the comparison or baseline, and the time and setting covered by the evidence. Then state assumptions that might connect the observed result to the broader claim, such as whether the comparison is fair or whether the result applies to other groups. NIST’s framework describes trustworthiness as contextual and socio-technical, so a technical measure alone may not settle a social claim.

Common mix-up: a system-level performance measure does not automatically establish a fair or beneficial social outcome.

• AI Risk Management Framework FAQs

  A concise guide to the framework’s purpose and when its characteristics apply.

AI systems are adopted, governed, and used by people and organizations. When assessing a public decision, ask who chose the system, who sets its purpose, who can challenge an outcome, and who is responsible for reviewing harms. The OECD principles call for transparency and safeguards, including human agency and oversight, and say information should help people adversely affected by an AI system challenge its output. NIST’s AI Risk Management Framework is voluntary guidance intended to help developers, users, and evaluators manage risks that may affect individuals, organizations, society, or the environment. These sources offer ways to structure questions; a framework or principle alone does not show that a particular institution followed it or achieved a good result.

For specific domains, continue to AI in education and research, AI at work, or AI media and information.

• AI Risk Management Framework FAQs

  The FAQs explain the framework’s voluntary purpose and intended users.

Ask how access to an AI-enabled service, its benefits, and its burdens are distributed. Who can use it? Who may be excluded or face barriers? Are outcomes different across groups, and what evidence would reveal that? OECD principles connect responsible AI stewardship with inclusion and reducing inequalities, while also calling for respect for non-discrimination and equality. These statements help identify issues to investigate; they do not establish that every AI deployment widens or narrows inequality. Evidence needs to fit the system, population, and setting under discussion.

For focused examples, explore AI in education and research, AI environmental impact, and AI at work.

• AI Risks and Trustworthiness

  The NIST framework describes trustworthiness characteristics as dependent on context and social and organizational factors.