From Moral Questioning to Moral Action: What I Learned About Technology, Ethics, and Change

Over seven weeks, I participated in Stanford University's McCoy Family Center for Ethics in Society "Ethics, Technology, and Public Policy for Practitioners" course. As Executive Director of the World Jewish Congress Technology and Human Rights Institute (TecHRI), I sought frameworks to navigate AI's ethical challenges to human dignity and democratic governance. What I found was more disruptive: it changed how I think about what's possible when we refuse to accept that technology's trajectory is predetermined.

A question posed has stayed with me: How do we go from moral questioning to moral action? In human rights work, we're trained to identify injustice. But with technology changing so fast, it's harder to know how to fix problems. This course gave me better tools for asking the right questions and the confidence to try new solutions.

Everything Is a Policy Choice—Including What We Build and When

One insight came from our session on AI and jobs: whether AI substitutes human workers or augments them is a human choice, not a market choice. This seems obvious, but it's often forgotten.

Here's what really matters: Product design is a policy choice. When to launch a product is a policy choice. Who to hire is a policy choice. Every policy choice involves tradeoffs—speed versus safety, profit versus people, innovation versus fairness. When we pretend these are just technical or business decisions, we hide the values behind each choice and avoid taking responsibility.

This matters for human rights work. Companies often claim they can't control how their technology affects people, that it's just "market forces." But if job displacement is a choice, then so is algorithmic bias and platform design that spreads hate. Jewish communities worldwide have been targeted by AI-generated hate speech, deepfake videos spreading antisemitic lies, and algorithms that amplify conspiracy theories. These aren't accidents—they're the results of specific choices about design, content rules, and business models.

This raises an uncomfortable question: Can any company really say they didn't see harm coming? When OpenAI released Sora, was anyone genuinely surprised that people immediately used it to create fake pornography and disinformation? The pattern is clear by now. Companies need to address predictable harms before launch, not after communities are already hurt.

The Challenge of Persuasion: How Do We Create Change Without Forcing It?

One tension that emerged throughout the course was this: how do we advocate for change without imposing our views on others? Real political change needs mass persuasion, not just regulations or moral authority.

This is especially hard when different groups have differing values. How do we build support for ethical AI when stakeholders have competing visions of what's right?

The course pushed me to think beyond traditional human rights tactics—calling out bad actors, filing legal complaints, citing international standards. Instead, we need strategies that show people how their own interests align with ethical outcomes. This means proving the business case for fair design, showing how good AI creates value beyond just avoiding harm, and demonstrating how AI can spread power rather than concentrate it.

For TecHRI, this means continuing to work with tech companies as potential partners, not just targets. It means talking with engineers about their real constraints, with product managers facing impossible deadlines, and with executives balancing competing demands. Moral clarity is important, but so is understanding others' perspectives.

Rethinking Regulation: We Need More Creative Tools

One surprising moment for me came when a speaker talked about using taxes as tech regulation. This challenges the usual "regulate or don't regulate" debate. Research shows that if we want AI that helps rather than replaces humans, we need policies that make that approach more profitable.

But we also need to ask tougher questions about regulation. Should regulators focus so heavily on transparency instead of safety? Transparency helps—we can't fix what we can't see—but it's not enough. An algorithm that discriminates is still harmful even if we can see how it works. We need better standards for whether AI actually works: Does this system do what it claims? How does it fail?

We need many different tools for dealing with AI harms. Some problems need impact reviews before launch. Others need ongoing audits. Some need real human review of important decisions. Others might work better with insurance requirements or liability rules.

Most importantly, we need what one speaker (who I wish I could name – but Chatham House rules) called the next stage of software engineering: societal assurance. Just as engineers now routinely think about security and accessibility, they need to think about social impact. What power imbalances does this system create or fix? Whose voices get heard or ignored?

This isn't about making engineers solve impossible moral puzzles. It's about building standard practices—impact reviews, diverse teams, clear accountability—that make thinking about society as routine as checking code.

Building for the Long Term

The course concluded with an insight quoting Toni Morrison: we overestimate the change we can make in one year but underestimate the change that can happen in ten years.

I reflected on my 9 years in the field of tech policy and online harms, and 1.5 years since establishing WJC TecHRI, and looking at the entire timeline, I am proud of some of what we have achieved and look forward to adjusting some of the work based on what I learned:

First, Broadening regulatory advocacy. While combating online antisemitism remains critical, we can engage with structural approaches to AI governance—building partnerships with tech workers, software engineers, and others.

Second, system-level thinking. We'll work upstream—advocating for impact assessments, better standards for whether AI works, and human rights due diligence before high-risk AI deployment.

Third, persuasion and coalition-building. The ten-year view requires making the affirmative case for beneficial AI—systems designed to augment human capability, democratize opportunity, and strengthen democratic institutions.

An Invitation to Work Together

Ursula Le Guin's "The Ones Who Walk Away from Omelas" opened this course. N.K. Jemisin's "The Ones Who Stay and Fight" closed it. We can't walk away from technology's harms, and we can't accept them as the cost of progress. We need to stay and fight—creatively, together, and with clear moral purpose.

I want to talk with others working on these questions: How do we build social impact into engineering practice? How do we make human-centered AI economically smart? How do we persuade rather than force, while still protecting vulnerable communities? And how do we make sure AI spreads power rather than concentrates it?

If you work on technology, policy, ethics, or human rights, I'd love to hear from you. Together, we can push technology toward justice.

What would it look like if we actually made that choice?

I'm grateful to McCoy Family Center for Ethics in Society, all the speakers (Chatham House rules) and my cohort members who made this learning journey possible. And my husband who took care of the kids at the most challenging of bedtime hours.

*The thoughts are my own, but GAI helped me articulate them better!