Share
Responsible and Ethical AI Frameworks
AI is popping up everywhere these days, and it's important to make sure we're using it responsibly. In this presentation we'll dive into how good data and processes play a huge role in making AI ethical and fair.
We'll chat about frameworks that help to collect, manage, and govern data to keep AI systems transparent and to mitigate bias. You'll hear about real-world examples that show the best ways to create AI that's accountable and trustworthy.
Plus, we'll look at what's happening around the world with responsible AI. It reviews ethical frameworks designed to address challenges such as bias, fairness, transparency, accountability, and privacy. Key frameworks discussed include the EU's Ethics Guidelines for Trustworthy AI, the IEEE's Ethically Aligned Design, and the AI Ethics Guidelines from the OECD. This global view will highlight why it's crucial for countries to work together and set common standards to tackle AI's ethical challenges.
Join us for a fun session that will give you a taste of responsible AI frameworks and why they're key to building a reliable and inclusive AI future.
Karen Lopez
Transcript
41 slides
-
Karen Lopez Microsoft MVP, Data Platform Microsoft Certified Trainer, vExpert
Data management expert, space enthusiast, and #TeamData evangelist www.datamodel.com @datachick.bksy.social -
Bjarni Valdimar Tryggvason Icelandic-born Canadian engineer and a NRC/CSA astronaut.
He served as a Payload Specialist on Space Shuttle mission STS-85 in 1997, a nearly 12-day mission to study changes in the Earth's atmosphere. Bjarni is the first, and as of 2024, only Canadian astronaut of Icelandic birth. https://en.wikipedia.org/w/index.php?title=Bjarni_Tryggvason&oldid=1289917372 -
Key Takeaways Explainable Trustworthy Safety & Security Transparent Human Focused
Metrics and Tools Monitoring Managing Risk Fairness -
TLAs…xLAs Principles • FAST • GREATE PLEA Prompt Design •
TRACI • CREATE Others • HITL • RLHF • RAG • XAI 8 (c) InfoAdvisors, Inc. -
Letter Core Principle Explanation F Fairness & Inclusivity Avoiding bias,
ensuring equity across demographics, and addressing systemic inequalities in data and outcomes. A Accountability & Auditability Assigning clear responsibility for AI decisions, enabling auditability, and ensuring recourse mechanisms. S Sustainability & Safety Minimizing environmental impact, ensuring long-term viability, and supporting social sustainability. T Transparency & Tracebility Making AI systems explainable, understandable, and open to scrutiny by stakeholders. 9 (c) InfoAdvisors, Inc. -
Letter Principle Explanation G Governance Establishing policies and oversight to
ensure AI aligns with ethical standards. R Reliability Ensuring AI systems consistently produce accurate and dependable outcomes. E Explainability Making AI decisions transparent so users understand how they work. A Accountability Holding developers and organizations responsible for AI decisions and consequences. T Transparency Clearly communicating AI’s capabilities, limitations, and data usage. P Privacy Protecting user data and ensuring confidentiality. L Lawfulness Adhering to legal and regulatory requirements in AI development. E Equity Preventing biases and ensuring fair AI treatment for all individuals. A Accessibility Designing AI systems to be inclusive and usable for diverse populations. 10 (c) InfoAdvisors, Inc. -
AI Frameworks and Guides Just a small sample 11 (c)
InfoAdvisors, Inc. -
IEEE Ethically Aligned Design https://sagroups.ieee.org/global-initiative/wp-content/uploads/sites/542/2023/01/ead1e.pdf 12 Universal Human Values Political
Self-Determination & Data Agency Technical Dependebility (c) InfoAdvisors, Inc. -
IEEE Ethically Aligned Design: Principles A/IS shall be created and
operated to respect, promote, and protect internationally recognized human rights. Human Rights A/IS creators shall adopt increased human well-being as a primary success criterion for development. Well-being A/IS creators shall empower individuals with the ability to access and securely share their data, to maintain people’s capacity to have control over their identity. Data Agency A/IS creators and operators shall provide evidence of the effectiveness and fitness purpose of A/IS. Effectiveness The basis of a particular A/IS decision should always be discoverable. Transparency A/IS shall be created and operated to provide an unambiguous rationale for all decisions made. Accountability A/IS creators shall guard against all potential misuses and risks of A/IS in operation. Awareness of Misuse A/IS creators shall specify and operators shall adhere to the knowledge and skill required for safe and effective operation. Competence 13 (c) InfoAdvisors, Inc. -
NIST AI Risk Management Framework (AIRMF) Core Functions Govern Establish
policies, procedures, and structures to manage AI risks and promote accountability. Map Understand the context, goals, and potential impacts of AI systems. Measure Assess and monitor AI system performance, risks, and trustworthiness characteristics. Manage Prioritize and respond to risks through mitigation strategies and continuous improvement. https://www.nist.gov/itl/ai-risk-management-framework 1 4 ( C ) I N F O A D V I S O R S , I N C . -
AI RMF Key Principles Human centricity Fairness Transparency Sustainability Privacy
enhancement Security and resilience 1 5 ( C ) I N F O A D V I S O R S , I N C . -
OECD AI Principles Inclusive growth, sustainabile development, well- being Human
rights, rule of law, democratice values, fairness, privacy Transparency and explainability Robustness, security, and safety Accountability, risk management 17 (c) InfoAdvisors, Inc. -
OECD AI Recommendations for Policy makers 18 Investing in AI
research and development Fostering an inclusive AI-enabling ecosystem Shaping an enabling interoperable governance and policy environment for AI Building human capacity and preparing for labour market transition International co-operation for trustworthy AI (c) InfoAdvisors, Inc. -
Canadian AI and Data Act (Proposed) Principle/Objective Description Protection from
Harm Prevent individual and collective harms caused by AI systems, including biased outputs and safety risks. Responsible Innovation Encourage AI development aligned with Canadian values like fairness, human rights, and democracy. Transparency and Oversight Require documentation and assessments of high-impact AI systems. Accountability Mandate risk management programs and designate responsible individuals for AI governance. Prohibition of Harmful Practices Criminalize deceptive or harmful AI uses (e.g., deepfakes, unauthorized biometric surveillance). International Alignment Align with global standards (e.g., EU AI Act) to ensure interoperability and trust. 19 (c) InfoAdvisors, Inc. -
Voluntary Code of Conduct on the Responsible Development and Management
of Advanced Generative AI Systems 20 Principle Description Accountability Organizations understand their role with regard to the systems they develop or manage, put in place appropriate risk management systems Safety Design and test AI systems to minimize risks of harm to individuals and society. Risk assessment completed Fairness and Equity Identify and mitigate bias to prevent discrimination against individuals or groups. Transparency Provide clear information about how AI systems work, their limitations, and their use. Ensure users can make informed decisions and experts can assess risks. Human Oversight Ensure meaningful human oversight, especially in high-impact or sensitive applications, including monitoring is deployed and use. Validity and Robustness Test AI systems for reliability, accuracy, and resilience to misuse or adversarial inputs. https://ised-isde.canada.ca/site/ised/en/voluntary-code-conduct-responsible-development-and-management-advanced-generative-ai-systems (c) InfoAdvisors, Inc. -
EU Ethics Guidelines for Trustworthy AI AI systems should empower
us to make informed decisions, and foster fundamental rights. Oversight needs to be ensured with human-in-the-loop, human-on-the-loop, and human-in- command approaches Human agency and oversight AI systems need to be resilient, secure, safe, accurate, reliable, and reproducible. Unintentional harm should be minimized and prevented. Technical Robustness and safety Ensuring privacy and data protection, data governance, data quality, integrity, and legitimised access to data. Privacy and data governance: Data, system and AI business models should be transparent and traceable, AI systems and decisions should be explainable. AI systems, capabilities and limitation should be known Transparency Unfair bias must be avoided. AI systems should be accessible to all, regardless of any disability, and involve relevant stakeholders throughout their entire life circle. Diversity, non- discrimination and fairness AI systems should benefit all human beings, including future generations. They should be sustainable and environmentally friendly. Societal and environmental well- being Ensure responsibility, auditability, and accountability for AI systems. Adequate an accessible redress should be ensured. Accountability 21 https://digital-strategy.ec.europa.eu/en/library/ethics-guidelines-trustworthy-ai (c) InfoAdvisors, Inc. -
EU AI Act The AI Act sets out a clear
set of risk- based rules for AI developers and deployers regarding specific uses of AI. The AI Act is part of a wider package of policy measures to support the development of trustworthy AI, which also includes the AI Innovation Package, the launch of AI Factories and the Coordinated Plan on AI. Together, these measures guarantee safety, fundamental rights and human-centric AI, and strengthen uptake, investment and innovation in AI across the EU. 22 (c) InfoAdvisors, Inc. -
Google AI Principles Bold Innovation Responsible development & deployment Collaborative
progress, together 23 https://ai.google/principles/. (c) InfoAdvisors, Inc. -
Google Responsible Generative AI Toolkit Design a responsible approach Align
your models Evaluate your models and system for safety Protect your system with safeguards 24 (c) InfoAdvisors, Inc. -
AWS Responsible AI Fairness Considering impacts on different groups of
stakeholders Explainability Understanding and evaluating system outputs Privacy and security Appropriately obtaining, using, and protecting data and models Safety Preventing harmful system output and misuse Controllability Having mechanisms to monitor and steer AI system behavior Veracity & robustness Achieving correct system outputs, even with unexpected or adversarial inputs Governance Incorporating best practices into the AI supply chain, including providers and deployers Transparency Enabling stakeholders to make informed choices about their engagement with an AI system 26 https://aws.amazon.com/ai/responsible-ai/ (c) InfoAdvisors, Inc. -
Framework Transparency & Explainability Fairness & Non- Discrimination Accountability Data
Privacy & Protection Human Oversight Inclusivity & Cultural Sensitivity Environmental Sustainability Ethical Data Sourcing Labor Market Impact OECD ✓ ✓ ✓ ✓ ✓ ✓ ✓ EU ✓ ✓ ✓ ✓ ✓ Singapore ✓ ✓ ✓ ✓ ✓ Australia ✓ ✓ ✓ ✓ ✓ Canada (AIDA) ✓ ✓ ✓ ✓ ✓ Mexico (UNESCO) ✓ ✓ ✓ ✓ ✓ UAE ✓ ✓ ✓ ✓ ✓ Saudi Arabia ✓ ✓ ✓ ✓ ✓ India ✓ ✓ ✓ ✓ ✓ ✓ Brazil ✓ ✓ ✓ ✓ ✓ Microsoft ✓ ✓ ✓ ✓ ✓ Google ✓ ✓ ✓ ✓ ✓ AWS ✓ ✓ ✓ ✓ ✓ African Observatory ✓ ✓ ✓ ✓ ✓ ✓ ✓ CIPIT ✓ ✓ ✓ ✓ ✓ ✓ ✓ 27 (c) InfoAdvisors, Inc. -
Tools for Responsible AI Just a small sample 28 (c)
InfoAdvisors, Inc. -
AWS Well-Architected Responsible AI Resources AWS Responsible AI Policy Responsible
AI Best Practices: Promoting Responsible and Trustworthy AI Systems Amazon AI Fairness and Explainability Whitepaper AWS generative AI Best Practices Framework AWS Responsible AI Landing Page 30 (c) InfoAdvisors, Inc. -
Guidelines for Human-AI Interaction HAX Workbook HAX Design Patterns HAX
Playbook HAX Design Library 32 (c) InfoAdvisors, Inc. -
OECD Tools & Metrics for Trustworthy AI Database https://oecd.ai/en/ 34
(c) InfoAdvisors, Inc. -
Data AI Readiness Invest in Data Governance Integrate Data Governance
with AI Governance Improve Data Quality processes Strengthen Data Literacy Integrate AI with Data Analytics 36 (c) InfoAdvisors, Inc. -
Data AI Readiness Refocus on metadata programs Strengthen security measures
Learn & Leverage AI for Data Managment Build a data-driven culture Build ethical AI skills, knowledge, and methods 37 (c) InfoAdvisors, Inc. -
More Frameworks for Ethical and Responsible AI 1.Asilomar AI Principles:
Developed during the 2017 Asilomar conference, these principles focus on research issues, ethics and values, and longer-term issues 2.Montreal Declaration for Responsible AI: A framework that outlines principles for responsible AI development, including well-being, respect for autonomy, and democratic participation 3.AI4People’s Ethical Framework for a Good AI Society: Offers recommendations and outlines principles for the ethical implementation of AI in society 38 (c) InfoAdvisors, Inc. -
More Frameworks for Ethical and Responsible AI 4. The Toronto
Declaration: A declaration focusing on protecting the right to equality and non-discrimination in machine learning system 5. The Beijing AI Principles: Developed by the Beijing Academy of Artificial Intelligence, these principles focus on harmony and friendliness, fairness and justice, and inclusivity and sharing 6. The Public Voice’s Universal Guidelines for AI: A set of principles that include the right to transparency, the right to human determination, and the right to redress2. 39 (c) InfoAdvisors, Inc.