This week, I had the privilege of presenting at the Global Data Summit in Reykjavik — a city known for its natural wonders, but also a strong tech industry. My session, Updating Data Programs with Responsible and Ethical AI focused on the intersection of data governance and artificial intelligence, and how organizations can prepare their data governance and data management programs for the ethical challenges and opportunities AI presents. This session was an excellent opportunity to discuss AI Ethics in a global context. This was my follow-up talk after presenting Responsible and Ethical AI Frameworks: An Introduction.
Why This Matters
AI is already embedded in our daily operations. But with great power comes great responsibility. As data professionals, we must ensure that our systems are not only efficient but also fair, accountable, sustainable, and transparent. These FAST ethical AI principles apply everywhere so we should ensure they are the focus of our AI uses in data management, adhering to AI Ethics at every step.
https://speakerdeck.com/datachick/updating-data-programs-with-responsible-and-ethical-ai
Lessons from the Field
I shared several cautionary tales that underscore the importance of responsible AI:
- Amazon’s biased hiring algorithm penalized resumes with the word “women,” revealing how historical data can perpetuate systemic bias. AI Ethics mandates that we monitor and fix bias issues.
- IBM Watson for Oncology struggled due to reliance on synthetic data, highlighting the risks of poor data quality in high-stakes environments. We need to ensure our test data is FAST and ethically sound.
- Air Canada’s chatbot fail showed the legal and reputational risks of deploying AI without clear accountability. They tried to claim that the chatbot was a separate legal entity. I’m glad the court disagreed on that, which is a win for AI Ethics.
- Zillow’s $380M USD loss due to “concept drift” in its pricing models reminded us that we must monitor and manage AI systems or risk costly consequences.
Building AI-Ready Data Programs
To responsibly integrate AI into data programs, I emphasized the need to:
- Invest in Data Governance: Align AI governance with existing frameworks, reflecting AI Ethics principles.
- Improve Data Quality: Garbage in, garbage out still applies.
- Implement Bias Detection: Monitor and mitigate bias continuously.
- Document Everything: Transparency starts with traceability.
- Use Synthetic Data Wisely: I’d even say that synthetic data is someone’s production data.
People and Process: The Real Challenge
Technology is only part of the equation. Change fatigue, fear of job loss, and ethical concerns are real barriers. That’s why I recommend teams ensure they are doing the following”
Implementing co-piloting approaches to ease adoption. I’m not talking about Microsoft Copilot here, but that the use of AI in our jobs mandates that we must see it as a helper, not a outsourcer. We should treat use similar to paired-programming. That means validating the inputs, outputs, and summaries. Every time.
Ramping up training programs focused on AI trust and literacy. This includes both AI and data literacy programs, especially focused on our own organization’s data and technical debt, with a key focus on AI Ethics.
Ensuring human-in-the-loop (HITL) processes to maintain integrity of our work. Today, I would not just set an AI helper out there to classify the data with no humans involved. That’s how we find ourselves in messes like confusing biodiversification with societal diversification.
Questions to Consider
1. How is your organization ensuring that AI systems are explainable and accountable to both internal and external stakeholders?
2. What steps are you taking to integrate ethical frameworks into your data governance and AI development processes?
3. Who in your data group is the single point of contact for AI-related questions and challenges?
4. Have our data programs been updated to both support the use of AI as well as support data uses for AI?
Calls to Action
Start a conversation within your team about how your current data practices align with ethical AI principles.
Commit to continuous learning whether through formal training or peer discussion to stay ahead of AI’s increasing impact on data programs, focusing on AI Ethics.
What are you doing to prepare yourself for using AI today?
One response
Of course I mean besides reading my content! 😁