ISO/IEC INTERNATIONAL STANDARD 22989 First edition 2022-07 Information technology Artificial intelligence Artificial intelligence concepts and terminology Technologies de I'information - Intelligence artificielle - Concepts et terminologie relatifs a I'intelligence artificielle Reference number EC ISO/IEC 22989:2022(E) ISO @ IS0/IEC 2022 IS0/IEC 22989:2022(E) COPYRIGHT PROTECTED DOCUMENT @IS0/IEC2022 All rights reserved. Unless otherwise specified, or required in the context of its implementation, no part of this publication may be reproduced or utilized otherwise in any form or by any means, electronic or mechanical, including photocopying, or posting on the internet or an intranet, without prior written permission. Permission can be requested from either Iso at the address below or Iso's memberbody inthe country of the requester. ISO copyright office CP 401 · Ch. de Blandonnet 8 CH-1214 Vernier, Geneva Phone: +41 22 749 0111 Email: [email protected] Website: www.iso.org Published in Switzerland ii @ IS0/IEC 2022 - All rights reserved IS0/IEC 22989:2022(E) Contents Page Foreword vi Introduction .vii 1 Scope 2 Normative references 3 Terms and definitions .1 3.1 Terms related to AI. .1 3.2 Terms related to data. 6 3.3 Terms related to machine learning .8 3.4 Terms related to neural networks 10 3.5 Terms related to trustworthiness 11 3.6 Terms related to natural language processing 13 3.7 Terms related to computer vision 16 4 Abbreviatedterms .16 5 AI concepts 17 5.1 General. 17 5.2 From strong and weak AI to general and narrow AI 17 5.3 Agent 17 5.4 Knowledge 18 5.5 Cognition and cognitive computing 19 5.6 Semantic computing. 19 5.7 Soft computing 19 5.8 Genetic algorithms. 19 5.9 Symbolic and subsymbolic approaches for Al 19 5.10 Data 20 5.11 Machine learning concepts 21 5.11.1 Supervised machine learning 21 5.11.2 Unsupervised machine learning 21 5.11.3 Semi-supervised machine learning 22 5.11.4 Reinforcement learning 22 5.11.5 Transfer learning 22 5.11.6 Training data. 22 5.11.7 Trained model 22 5.11.8 Validation and test data 22 5.11.9 Retraining. 23 5.12 Examples of machine learning algorithms 24 5.12.1 Neuralnetworks 24 5.12.2 Bayesian networks 25 5.12.3 Decision trees.. 25 5.12.4 Support vector machine. 25 5.13 Autonomy, heteronomy and automation. 26 5.14 Internet of things and cyber-physical systems 27 5.14.1 General 27 5.14.2 Internet of things 27 5.14.3 Cyber-physical systems. 27 5.15 Trustworthiness 28 5.15.1 General 28 5.15.2 AI robustness. 28 5.15.3 AI reliability 29 5.15.4 AI resilience 29 5.15.5 AI controllability 29 5.15.6 AI explainability 29 5.15.7 AI predictability 30 @ IS0/IEC 2022 - All rights reserved ili IS0/IEC 22989:2022(E) 5.15.8 AI transparency .30 5.15.9 AI bias and fairness 30 5.16 AI verification and validation. 31 5.17 Jurisdictionalissues. 31 5.18 Societal impact 32 5.19 AI stakeholder roles. .32 5.19.1 General 32 5.19.2 AI provider. .33 5.19.3 AI producer. .33 5.19.4 AI customer. 34 5.19.5 AI partner .34 5.19.6 AI subject. .34 5.19.7 Relevant authorities .35 6 AI system life cycle .35 6.1 AI system life cycle model .35 6.2 AI system life cycle stages and processes .37 6.2.1 General 37 6.2.2 Inception .37 6.2.3 Design and development 38 6.2.4 Verification and Validation 39 6.2.5 Deployment. .39 6.2.6 Operation and monitoring .39 6.2.7 Continuous validation. .40 6.2.8 Re-evaluation 40 6.2.9 Retirement 40 7 Al system functional overview 40 7.1 General .40 7.2 Data and information. 41 7.3 Knowledge and learning 41 7.4 From predictions to actions 42 7.4.1 General. 42 7.4.2 Prediction 42 7.4.3 Decision. 43 7.4.4 Action 43 8 AI ecosystem .43 8.1 General. 43 8.2 AI systems. 45 8.3 AI function. 45 8.4 Machine learning 45 8.4.1 General .45 8.5 Engineering. .46 8.5.1 General 46 8.5.2 Expert systems .46 8.5.3 Logic programming 46 8.6 Big data and data sources cloud and edge computing 46 8.6.1 Big data and data sources. 46 8.6.2 Cloud and edge computing 48 8.7 Resource pools .50 8.7.1 General .50 8.7.2