Artificial intelligence can analyse information, identify patterns and generate outputs at remarkable speed. But when AI-supported test results inform important decisions, speed and sophistication are not enough.
The question that matters is: why should we trust it?
As AI moves from experimentation into everyday use, organisations are confronting questions about reliability, transparency, data quality, governance and accountability. These questions become particularly important when results may affect health, safety, trade, regulatory decisions or public confidence.
An AI system may appear highly capable. Capability alone, however, does not tell us whether its data is appropriate, its limitations are understood or its output is suitable for its intended purpose.
To adopt AI confidently, organisations need more than assurances about what the technology can do. They need credible evidence that it is being governed, evaluated and used responsibly.
Trust needs evidence
In her article, Trust is the infrastructure of scale, Microsoft standards strategist Laura Lindsay argues that trust will determine whether emerging technologies can move from breakthroughs to widespread use.
An AI system can produce an impressive output without giving the user sufficient confidence to act on it. If a result appears wrong, can it be questioned? Is the information it relies on suitable? Are its limitations understood? Is a competent person responsible for reviewing the outcome?
These are not arguments against AI. They are the questions responsible adoption requires.
For the accreditation community, the underlying principle is familiar: important claims should be demonstrated, not simply asserted.
Standards establish agreed requirements. Conformity assessment is the mechanism that determines whether those requirements have been fulfilled.
Accreditation provides formal recognition of the competence of organisations performing specific conformity assessment activities, but conformity assessment is only meaningful when it’s performed by competent persons and organisations
Together, these elements help regulators, businesses and the community place greater confidence in results.
Standards are building a foundation
ISO/IEC 42001:2023 is the first international management system standard for artificial intelligence. It provides a structured approach for organisations to manage the risks and opportunities associated with developing, providing or using AI systems.
Importantly, it addresses organisational governance. It does not certify that an individual AI application will always be accurate, safe or appropriate. Organisations must still consider how a system performs in its intended context.
Before relying on an AI-supported result
If AI is introduced into a technical process, organisations should consider:
- Purpose: What task is the system being used to perform?
- Data: Is the information it relies on appropriate, reliable and controlled?
- Validation: Has its performance been evaluated for its intended use?
- Limitations: Do users understand when it may produce unreliable results?
- Oversight: Is a competent person reviewing its outputs?
- Accountability: Is it clear who remains responsible for the final result?
How these considerations apply will depend on the technology, its intended use and whether it could influence an accredited activity or outcome. They should not be treated as a formal NATA checklist, but as useful questions to support responsible decision-making. In fact, the principles for these considerations are already in existing accreditation standards like ISO/IEC 17025.
AI-generated information should be treated as an input to be evaluated, not as an unquestionable conclusion.
What does this mean for NATA and accredited facilities?
NATA has identified verification and validation, control of data, organisational governance and accountability as areas of increasing importance as AI and automated systems become more widely adopted.
For accredited facilities, the critical issue is not simply whether AI is being used. It is whether that use could influence an accredited activity or outcome.
The fundamentals of confidence remain unchanged. Facilities still need competent people, appropriate methods, controlled processes and valid results. Where AI is involved, they must understand how it is being used, recognise its limitations and retain responsibility for the outcome.
NATA is also using ISO/IEC 42001 as a guiding framework for responsible AI use within its own organisation. Its approach includes reviewing AI-generated content, verifying information, protecting sensitive information and applying human judgement.
For nearly 80 years and across all sectors of the economy, NATA has provided confidence in the competence of organisations who perform conformity assessment activities. Although the technologies have evolved, the principles of our independent 3rd party accreditation processes remain unchanged, and we will continue to provide confidence in the age of AI.
Trust still has to be earned. AI can support a result. It cannot take responsibility for it.