AI in the energy sector guidance consultation
Appendix 2: AI standards
2.1 AI is a nascent technology, and standards and frameworks have been and continue to be developed to support the ethical use of AI. Standards are agreed ways of doing something. These cover, for example, how to make a product, manage a process, or deliver a service. Standards are open, consensus-based guidance that represent good practice. Standards are produced by national and international standards bodies (such as the British Standards Institute in the UK and the International Organisation of Standardisation). These product and quality standards are outside of Ofgem’s remit but can be used by stakeholders to manage risks associated with AI use. Ofgem may take adherence to standards and frameworks into account when assessing compliance with this guidance.
2.2 This appendix identifies how selected AI standards, whether published or in draft form, and key frameworks map onto the government’s five AI principles.
2.3 Read the standards listed on the AI Standards Hub.
AI standards mapped to the five AI principles
|
Standards with principle as core focus (core and related) |
Safety, security, and robustness |
Appropriate transparency and explainability |
Fairness |
Accountability and governance |
Contestability and redress |
|
Published |
12 |
8 |
3 |
8 |
0 |
Source: British Standards Institute, October 2024
Standards: published
|
|
Safety, security, and robustness |
Appropriate transparency and explainability |
Fairness |
Accountability and governance |
Contestability and redress |
|
Artificial intelligence management system |
Yes |
Yes |
Yes |
Yes (core focus) |
No |
|
Safety of machinery Relationship with ISO 12100 Part 5: Implications of artificial intelligence machine learning |
Yes (core focus) |
No |
No |
No |
No |
|
Bias in AI systems and AI aided decision making |
Yes |
Yes |
Yes (core focus) |
Yes |
No |
|
Assessment of the robustness of neural networks |
Yes (core focus) |
No |
No |
No |
No |
|
Overview of trustworthiness in artificial intelligence |
Yes |
Yes (core focus) |
Yes |
No |
No |
|
Governance implications of the use of artificial intelligence by organisations |
No |
No |
No |
Yes (core focus) |
No |
|
Controllability of automated artificial intelligence systems |
Yes |
No |
No |
No |
No |
|
Functional safety and AI systems |
Yes (core focus) |
No |
No |
No |
No |
|
Software and systems engineering: Software testing |
Yes (core focus) |
Yes |
No |
Yes |
No |
|
Governance of AI |
No |
Yes (core focus) |
No |
Yes |
No |
|
Systems and software Quality Requirements and Evaluation (SQuaRE) Quality model for AI systems |
Yes (core focus) |
Yes |
No |
Yes |
No |
|
Information technology. Governance of IT Governance implications of the use of artificial intelligence by organisations |
Yes |
Yes |
No |
Yes |
No |
|
ISO/IEC 23894:2023 AI Guidance Risk Management |
Yes |
No |
No |
No |
No |
|
AI management system standard |
Yes |
Yes |
No |
Yes |
No |
Source: British Standards Institute, October 2024
Standards: under development
a. ISO/IEC CD 12792 Transparency taxonomy of AI systems
b. ISO/IEC CD TS 6254 Objectives and approaches for explainability and interpretability of ML models and AI systems
c. ISO/IEC DTS 12791 Treatment of unwanted bias in classification and regression machine learning tasks
d. ISO/IEC AWI TS 17847 Verification and validation analysis of AI systems
e. ISO/IEC CD TR 20226 Environmental sustainability aspects of AI systems
f. ISO/IEC AWI TR 21221 Information technology and artificial intelligence: beneficial AI systems
g. ISO/IEC AWI TS 22443 Guidance on addressing societal concerns and ethical considerations
h. ISO/IEC AWI TS 22440 Functional safety and AI systems requirements
i. ISO/IEC AWI 23282 Evaluation methods for accurate natural language processing systems
j. ISO/IEC DIS 42005 Information technology and artificial intelligence: AI system impact assessment
k. ISO/IEC DIS 42006 Requirements for bodies providing audit and certification of artificial intelligence management systems
l. ISO/IEC AWI 42105 Guidance for human oversight of AI systems
m. ISO/IEC DIS 5259-5 Data quality for analytics and machine learning (ML) Part 5: data quality governance framework
n. ISO/IEC AWI TS 42119-2 Artificial intelligence testing of AI part 2: overview of testing AI systems
o. ISO/IEC 21031 Information technology Software carbon intensity (SCI) specification
Source: British Standards Institute, October 2024