Somewhere in the last two years, the conversation about AI shifted, and you must be questioning me here. “Honey, we know this. What is new in this article?”
Well, it used to be about possibility. What can it do? How fast can it go? What becomes possible that wasn't before? Those are still real questions. But increasingly, the question underneath all of them is starting to surface: What kind of world are we building with this?
I've spent decades in technology, building platforms for civic safety, environmental conservation, hunger relief, and governance. What I've learned, slowly and sometimes the hard way, is that technology is never neutral. Every system reflects a set of choices. And every set of choices reflects a set of values, whether the builder was conscious of them or not.
The scale at which AI now operates makes this more urgent than any previous moment in technology history. When a biased algorithm was running inside one company's hiring tool, the harm was contained. When an AI system operating on those same biases is deployed across thousands of companies simultaneously, the harm scales identically. The efficiency of AI applies to mistakes as efficiently as it applies to progress.
MIT Sloan researchers studying enterprise AI adoption found something striking: most organizations are adopting AI without first asking what their own values are. They have product roadmaps. They have deployment timelines. What they don't have is a clear answer to the question: What do we actually stand for, and does this system reflect that?
The failures that have emerged from AI deployments are rarely technical. The McDonald's AI hiring platform that exposed 64 million job application records wasn't brought down by a sophisticated attack; it had a default password of "123456" and no two-factor authentication. The lesson wasn't about AI. It was about governance. About who owns responsibility for these systems when they touch real people's lives.
Here's what I believe, for whatever it's worth after 25+ years of building technology with a social conscience:
The organizations that will earn lasting trust in the AI era are not those who move fastest. They are those who can answer a simple set of questions before they deploy: Who is accountable when this system is wrong? What values is this system optimizing for—and whose values are those? What do we do when the outcome is technically correct but ethically questionable?
These are not philosophical questions. They are operational ones. And right now, most organizations don't have answers.
The EU's proposed Digital Fairness Act, regulatory conversations happening globally, and the lawsuits now going to trial—these are not obstacles to AI. They are the world asking the question that builders should have been asking first. Society is filling the wisdom gap that the industry left open.
I don't think AI is the threat. I think the absence of deliberate thought about what we're building is.
The most powerful tools in history have always required the most careful hands. Fire, the printing press, and the internet each transformed the world, and each caused harm in proportion to how unconsciously it was deployed. AI is no different, except that it's faster, broader, and less transparent than anything that came before it.
We spent years perfecting the engine. It's time to talk seriously about where we're driving.
If your organization deployed an AI system tomorrow and it caused harm, who, specifically, is accountable? Not the vendor. Not the algorithm. A person. Do you know that person's name?