5 Practical Tips to Building AI Products Responsibly
Shyvee Shi
Product @ Microsoft | A forward-thinking product leader combining creativity, user psychology, and AI to drive growth and scale communities | ex-LinkedIn
Welcome to the latest issue of the Product Management Learning Series - a series of live streaming events and newsletter articles to help you level up your product career! 🚀
In our 34th installment, our speaker was Navrina Singh, Founder and CEO of Credo AI and a technology leader with over 18 years of experience in Enterprise SaaS, AI, and Mobile. Navrina is a member of the U.S. Department of Commerce National Artificial Intelligence Advisory Committee (NAIAC) and an executive board member of Mozilla guiding their trustworthy AI charter. She is also a young global leader with the World Economic Forum and was on their Future Council for AI guiding policies and regulations in responsible AI.
If you missed the event, you can watch the full event recording here.
Below are the main takeaways from my conversation with Navrina:
There are three key challenges to responsible AI: oversight deficiency, governance as a hindrance vs. enabler, and the techno-social impact of AI.
Navrina explained the concept of Responsible AI which is an emerging field that focuses on bringing accountability and oversight to the design, development, and use of artificial intelligence systems. With the rapid development of AI technology, responsible AI aims to ensure that the intended impact of these systems aligns with ethical values and regulatory frameworks. Operationalizing responsible AI involves governance, which requires companies and organizations to incorporate safety, fairness, security, compliance, auditability, and human-centered design into their AI development processes.
She then identified three key challenges of responsible AI: the oversight deficit, governance as a hindrance versus an enabler, and the techno-social impact of AI. The oversight deficit creates a challenge in bringing together multiple stakeholders, including risk and compliance functions and impacted communities, requiring a shift in thinking for product managers to bridge the gap between technical and non-technical viewpoints. Governance is often perceived as a hindrance to speed and innovation rather than an enabler, and it is necessary to change that perspective. The third challenge involves the techno-social implications of AI systems, such as fairness and adversarial implications, leaving certain demographics out because of biases in training data.
To compact these challenges in AI, product managers must upskill themselves and broaden their perspective to include regulations, demographic impacts, and engagement with business stakeholders and risk professionals.
Navrina Singh emphasized the importance of product managers upskilling in the field of AI, as machine learning transitions from being solely a technical problem to a technical social problem. While product managers already possess communication skills and the ability to bring in diverse perspectives, they need to consider regulations and potential demographic impacts previously overlooked when building products.
Navrina suggested ways product managers can upskill in AI, such as engaging with business stakeholders to understand their priorities and working closely with risk professionals to mitigate business regulatory brand risk. She also highlighted the importance of responsible AI governance, especially with generative AI in mind.
Product managers must keep learning and unlearning to remain relevant in the AI-powered world.
Product Managers need to continuously learn and unlearn in the field of AI, as the pace of advancements in AI is rapidly growing. Navrina shared that the best way to learn is to dive deeper into state-of-the-art papers that talk about the new AI advancements.
In terms of unlearning, it is critical for Product Managers to expand their stakeholder base. They need to understand where bias can emerge, and how unintended consequences for different demographics can be avoided. Navrina emphasized that engaging with customers and paying attention to how they are impacted is crucial. Product Managers should stay up to date on their learning methodologies and unlearn old product management behaviors and skills that don't serve in this new AI-powered world.
Companies struggled with implementing responsible AI due to a lack of standards, the emergence of regulations, siloed implementation of Machine Learning Operations pipelines, and lack of compliant data for testing and auditing.
According to Navrina, the advancement of AI is increasing the risk exposure of organizations, who are struggling to understand what responsible AI looks like. There is a lack of standards and benchmarks, which makes it difficult to measure performance and fairness in AI. However, there is an emergence of frameworks, regulations, and standards across the globe. The EU AI Act is a current example, which is focused on using risk categorization of AI to come up with a mechanism for conformity assessments and putting guardrails around those systems. The United States is also developing local and state regulations, such as New York City Law number 144, which requires an enterprise that is building or buying third-party AI employment systems to provide a fairness audit on their websites.
Another challenge is the siloed implementation of machine learning operations (ML Ops) pipelines. It is important to infuse accountability and oversight across all of them. There is also a lack of compliant data for testing and auditing. For instance, facial recognition systems have been highly debated due to the backlashes around how some of these systems don't work well on different demographics, especially women of color. Product managers need to start asking where the data is coming from, if there are governance structures in place, and how they can keep pace with emerging regulations.
Take intentional action and start the journey toward greatness today, even if you don't feel ready.
Navrina urged Product Managers to bias towards intentional action. In her view, waiting for perfection or greatness can hinder progress. Instead, she recommended taking intentional action and starting with small steps toward a goal. This mindset can help PMs navigate the complex and ever-changing AI landscape and bring about meaningful change. She emphasized the importance of starting the journey today and not waiting for the perfect time or conditions. The current macroeconomic situation makes it even more critical for PMs to take intentional action to bring intentional velocity to create positive change. The quote "you don't have to be great to get started, but you have to get started to be great" resonated with Navrina.
Additional gems from Navrina:
🍪 Navrina’s favorite product is GPT-4 from Open AI due to its massive implications and ease of use. However, she also cautioned about the unintended consequences that are already arising from its use, such as fairness challenges, privacy leakage, and toxicity issues. Navrina pointed out that it's a critical moment in the development of generative AI where the balance between ease of use and risk exposure needs to be carefully considered.
🍒 Navrina shared that she constantly learns from every customer interaction which has been one of the greatest product role models. Navrina believes that giving oneself the permission to be surprised is the best gift a product manager can give themselves.
💎 List of resources from Navrina:
Recommended by LinkedIn
[3] AI Standards Hub
🎉 Special kudos to Soumya Chaudhari for writing this article.
Next up, don’t miss out on an important live conversation we will have the very best Casey Winters , one of the leading experts in scaling startups and driving product growth!
👉 RSVP here (Replay available upon registration)
Next up,
More sessions coming… subscribe to stay tuned!
Learn more about the Product Management Learning Series and view past recordings here.
👩🏽💻Computer Engineer| 🎨 Creative Maverick | UI/UX Designer 🚀 | Figma Aficionado 🎨 | Web Designer Expert | Webflow Specialist | Cross-Functional Collaborator 🤝| Crafting Exceptional User Experiences 💡
1yAmazing!!!
Chief Executive Officer specializing in Business Operations and Data Science
1y😎 Thank you Shyvee for another great edition. Timely and insightful as always. 🙌
Sales Associate at Microsoft
1yGreat opportunity
Director of Platform Select/Striive
1yMaaike van Driel