Three lessons from ethics and public policy that changed how I think about AI

Building AI is not just about capability. It is about understanding fairness, amplifying human agency, and embracing responsibility before deployment.

For a long time, like many engineers, I believed that better performance naturally translated into better outcomes. If systems became more capable, efficient, and intelligent, society would benefit accordingly.

Experience — and later the Ethics, Technology + Public Policy for Practitioners program at Stanford University — challenged that assumption.

Technology does not operate in isolation. Questions about fairness, accountability, trust, and human values cannot simply be optimized away. They require judgment, humility, and an appreciation for complexity. Three ideas, in particular, transformed how I think about AI and technology.

---

Lesson One

Fairness is not something you optimize

One of the most influential moments for me came from discussions around algorithmic fairness. Reading Arvind Narayanan's "What If Algorithmic Fairness Is a Category Error?" alongside a hiring-algorithm case study challenged the way I thought about machine learning systems.

Until then, I often assumed fairness was another objective that could simply be optimized through better metrics. I gradually realized that questions of fairness involve competing values, historical context, and human judgment.

Metrics can reveal tradeoffs. They cannot eliminate them.

Metrics and models cannot eliminate difficult social choices — they only make those choices more visible.

Lesson Two

AI should amplify humans, not replace them

Another idea that stayed with me came from Susan Benesch's work on counterspeech and online civility. I was particularly struck by the notion that AI should amplify human efforts rather than replace them.

Thousands of people voluntarily promote civility and challenge hate speech online. Technology can support these efforts, but empathy, humor, trust, and civic participation remain deeply human qualities.

Technology should strengthen human agency — not substitute for it.

Lesson Three

Disagreement is not failure

One of the most valuable realizations I gained from the program was that disagreement should not be feared.

Complex technologies involve competing values. Different perspectives are not obstacles — they are part of responsible innovation. Progress emerges not from unanimity, but from thoughtful disagreement. Listening to disagreement is often how better decisions are made.

---

Ethics begins before deployment

Responsible systems are not created by adding ethics later. They are shaped by asking better questions before building.

Personal Statement of Practice

Submitted to the Stanford Ethics, Technology + Public Policy for Practitioners Program

---

Final thoughts

Perhaps the greatest gift of ethics and public policy is not certainty. It is learning to live comfortably with complexity.

Technology will continue to evolve. Capabilities will improve. Models will become increasingly powerful. But capability alone is not enough — the challenge before us is not merely to build smarter systems. It is to build systems worthy of trust.

I do not hope to be remembered simply for the technologies I helped create.

I hope to be remembered as someone who approached technology thoughtfully, listened carefully to others, and treated the responsibilities that accompany innovation with seriousness and care.