Read the original document by opening this link in a new tab.
Summary
Director General - Smart Dubai Office
RESPONSIBILITY
The Smart Dubai Office will not be responsible for any misuse of the AI Ethics Principles and Guidelines. The user bears all the consequences of their use.
LICENSING
This document is published under the terms of a Creative Commons Attribution 4.0 International Licence in order to facilitate its re-use by other governments and private sector organisations. In summary this means you are free to share and adapt the material, including for commercial purposes, provided that you give appropriate credit to the Smart Dubai Office as its owner and do not suggest the Smart Dubai Office endorses your use.
AI PRINCIPLES
AI should be beneficial to humans and aligned with human values, in both the long and short term
AI should benefit all people in society, be governed globally, and respect dignity and people rights
DUBAI’S AI PRINCIPLES
HUMANITY
INCLUSIVENESS
AI systems should be safe and secure, and should serve and protect humanity
SECURITY
AI systems should be fair, transparent, accountable and understandable
ETHICS
WE WILL MAKE AI SYSTEMS FAIR
- Data ingested should, where possible, be representative of the affected population
- Algorithms should avoid non-operational bias
- Steps should be taken to mitigate and disclose the biases inherent in datasets
- Significant decisions should be provably fair
WE WILL MAKE AI SYSTEMS ACCOUNTABLE
- Accountability for the outcomes of an AI system lies not with the system itself but is apportioned between those who design, develop and deploy it
- Developers should make efforts to mitigate the risks inherent in the systems they design
- AI systems should have built-in appeals procedures whereby users can challenge significant decisions
- AI systems should be developed by diverse teams which include experts in the area in which the system will be deployed
WE WILL MAKE AI SYSTEMS TRANSPARENT
- Developers should build systems whose failures can be traced and diagnosed
- People should be told when significant decisions about them are being made by AI
- Within the limits of privacy and the preservation of intellectual property, those who deploy AI systems should be transparent about the data and algorithms they use
AI PRINCIPLES
WE WILL MAKE AI SYSTEMS AS EXPLAINABLE AS TECHNICALLY POSSIBLE
- Decisions and methodologies of AI systems which have a significant effect on individuals should be explainable to them, to the extent permitted by available technology
- It should be possible to ascertain the key factors leading to any specific decision that could have a significant effect on an individual
- In the above situation we will provide channels through which people can request such explanations
SECURITY
AI SYSTEMS WILL BE SAFE, SECURE AND CONTROLLABLE BY HUMANS
- Safety and security of the people, be they operators, end-users or other parties, will be of paramount concern in the design of any AI system
- AI systems should be verifiably secure and controllable throughout their operational lifetime, to the extent permitted by technology
- The continued security and privacy of users should be considered when decommissioning AI systems
- AI systems that may directly impact people’s lives in a significant way should receive commensurate care in their designs, and;
- Such systems should be able to be overridden or their decisions reversed by designated people
AI SYSTEMS SHOULD NOT BE ABLE TO AUTONOMOUSLY HURT, DESTROY OR DECEIVE HUMANS
- AI systems should be built to serve and inform, and not to deceive and manipulate
- Nations should collaborate to avoid an arms race in lethal autonomous weapons, and such weapons should be tightly controlled
- Active cooperation should be pursued to avoid corner-cutting on safety standards
- Systems designed to inform significant decisions should do so impartially
WE WILL GIVE AI SYSTEMS HUMAN VALUES AND MAKE THEM BENEFICIAL TO SOCIETY
- Government will support the research of the beneficial use of AI
- AI should be developed to align with human values and contribute to human flourishing
- Stakeholders throughout society should be involved in the development of AI and its governance
WE WILL PLAN FOR A FUTURE IN WHICH AI SYSTEMS BECOME INCREASINGLY INTELLIGENT
- Governance models should be developed for artificial general intelligence (AGI) and superintelligence
- AGI and superintelligence, if developed, should serve humanity as a whole
- Long-term risks of AI should be identified and planned for
- Recursively self-improving AI development should be disclosed and tightly monitored and controlled for risk
HUMANITY
INCLUSIVENESS
AI PRINCIPLES
WE WILL GOVERN AI AS A GLOBAL EFFORT
- Global cooperation should be encouraged to ensure the safe governance of AI
- Government will support the establishment of internationally recognized standards and best practices in AI, and when they are established, shall adhere to them
WE WILL SHARE THE BENEFITS OF AI THROUGHOUT SOCIETY
- Development of AI systems will be matched by a response to its impact on employment
- AI will be used to help humans retain purpose and flourish mentally, emotionally and economically alongside AI
- Access to training, opportunity and tools should be made available
- Education should evolve and reflect the latest developments in AI, enabling people to adapt to societal change
WE WILL PROMOTE HUMAN VALUES, FREEDOM AND DIGNITY
- AI should improve society, and society should be consulted in a representative fashion to inform the development of AI
- Humanity should retain the power to govern itself and make the final decision, with AI in an assisting role
- AI systems should conform to international norms and standards with respect to human values and people rights and acceptable behavior
WE WILL RESPECT PEOPLE’S PRIVACY
- AI systems should respect privacy and use the minimum intrusion necessary
- AI systems should uphold high standards of data governance and security, protecting personal information
- Surveillance or other AI-driven technologies should not be deployed to the extent of violating internationally and/or UAE’s accepted standards of privacy and human dignity and people rights
CHANGELOG
AI PRINCIPLES VERSION STAGE DATE COMPLETED
SUMMARY OF CHANGES
1.0 1.1 1.2 1.3 1.4 1.5 1.6
Internal Internal Consultation Consultation Consultation Feedback Internal Review
06/09/2018 06/09/2018 10/09/2018 10/09/2018 25/09/2018 09/10/2018 30/12/2018
First draft Design Add elaboration under each principle Changes following steering meeting 2: Alignment to Humanity Changes following steering meeting 2 Equality to Inclusiveness Changes following first round of feedback by public and private sector entities Minor changes to the wording of principles: Humanity, Inclusiveness
This document was produced by Smart Dubai in collaboration with Falcon and Associates.
AI GUIDELINES
INTRODUCTION
AI’s rapid advancement and innovation potential across a range of fields is incredibly exciting. Yet a thorough and open discussion around AI ethics, and the principles organizations using this technology must consider, is urgently needed. Dubai’s Ethical AI Toolkit has been created to provide practical help across a city ecosystem. It supports industry, academia and individuals in understanding how AI systems can be used responsibly. It consists of principles and guidelines, and a self-assessment tool for developers to assess their platforms. The Dubai AI Ethics Guidelines relate to Ethics principle in Dubai AI Principles:
“AI systems should be fair, transparent, accountable and understandable”
They offer tangible suggestions to help stakeholders adhere to the principle. They deliver detailed guidance and are arranged according to the four sub-principles of Ethics principle:
• We will make AI systems fair
• We will make AI systems accountable
• We will make AI systems transparent
• We will make AI systems as explainable as technically possible
The guidelines are non-binding, and are being drafted as a collaborative, multi-stakeholder effort, with full awareness of organizations’ needs to innovate and protect their intellectual property. This is a collaborative process where all stakeholders are invited to be part of the dialogue. We would like to see the Dubai AI Ethics Guidelines evolve into a universal, practical and applicable framework informing ethical requirements for AI design and use.
With these guidelines our aim is to offer unified guidance that is continuously improved in collaboration with our communities. The eventual goal is to reach widespread agreement and adoption of commonly-agreed policies to inform the ethical use of AI not just in Dubai but around the world.
DUBAI AI ETHICS GUIDELINES
SCOPE
This document gives guidelines for achieving the ethical design and deployment of AI systems in both the public and private sectors. Specifically it covers the crucial issues of Fairness, Accountability, Transparency and Explainability of the algorithms at the heart of AI systems. This document does not cover issues relating to employment, security or any other aspects of the governance of artificial intelligence besides those mentioned above. AI already surrounds us, but some applications are more visible and sensitive than others. This document is applicable only to those AI systems which make or inform ‘significant decisions’ – that is, those decisions which have the potential for significant impact either on individuals or on society as a whole. They also apply to ‘critical decisions’, which are a subset of significant decisions and are of especially critical nature. See the Definitions section for a full definition of ‘significant decision’ and ‘critical decision’.
DEFINITIONS
For the purposes of these guidelines, the following definitions apply:
AI developer organization
An organization which does any of the following:
• determine the purpose of an AI system;
• design an AI system;
• build an AI system, or:
• perform technical maintenance or tuning on an AI system
Note 1 to entry: the definition applies regardless of whether the organization is the ultimate user of the system, or whether they sell it on or give it away
EXAMPLE: A company develops an artificially intelligent facial recognition system and sells it to a country’s border force, who use it to identify suspicious personnel. The company is an AI developer organization and the border force is an AI operator organization
AI OPERATOR ORGANIZATION
An organization which does any of the following:
• use AI systems in operations, backroom processes or decision-making;
• use an AI system to provide a service to an AI subject;
• is a business owner of an AI system;
• procure and treat data for use in an AI system, or:
• evaluate the use case for an AI system and decide whether to proceed
Note 1 to entry: this definition applies regardless of whether the AI system was developed in-house or procured. Note 2 to entry: it is possible for organizations to be both an AI developer organization and an AI operator organization
ARTIFICIAL INTELLIGENCE (also “AI”)
The capability of a functional unit to perform functions that are generally associated with human intelligence such as reasoning, learning and self-improvement1.
ARTIFICIALLY INTELLIGENT SYSTEM (also “AI system”)
A product, service, process or decision-making methodology whose operation or outcome is materially influenced by artificially intelligent functional units
Note 1 to entry: it is not necessary for a system’s outcome to be solely determined by artificially intelligent functional units in order for the system to be defined as an artificially intelligent system
Note 2 to entry: a particular feature of AI systems is that they learn behavior and rules not explicitly programmed in
EXAMPLE: A small claims court uses an artificially intelligent software package to collect evidence pertaining to a case, compare it to similar cases in the past, and present a recommended decision to a judge. The judge determines the final outcome. This decision-making methodology is materially influenced by an artificially intelligent functional unit, and is therefore classified as an AI system.
EXAMPLE: A government entity uses a chatbot which allows customers to ask routine questions, book appointments and conduct minor financial transactions. The chatbot responds to customer queries with pre-written responses and is based on pre-programmed decision rules. Therefore the chatbot is not an AI system. If, however, the chatbot autonomously adjusted its treatment of customers based on the outcome of past cases, it would be an AI system.
BIAS (of a system)
Inclination or prejudice for or against one person or group, especially in a way considered to be unfair2.
CRITICAL DECISION
An individually significant decision which is deemed to either have a very large impact on an individual or to have especially high stakes, be especially sensitive, have the potential to cause high loss or damage, is societally significant, or sets important precedent.
Note 1 to entry: the types of decisions referred to here are the same as those in the definition of significant-at-scale decisions, except in this case the effects are felt as a result of an individual decision rather than an aggregate of many decisions.
EXAMPLE: A court determines whether a defendant is guilty of a criminal charge, with the punishment for guilt being a life sentence. This is a critical decision because it has