Who Decides When AI Is Truly Ready to Act?

Jan de Vries · 2026-08-18

You know, we've been hearing a lot about AI lately, right? It's everywhere – in our phones, our cars, even helping doctors. And a big part of that conversation is about AI governance. We've gotten pretty good at spotting the potential dangers and mapping out all the risks involved with these super-smart systems. But here's a question that doesn't get asked enough: who actually gets to decide when an AI isn't just smart, but truly *ready* to make important decisions on its own? When does it have that legitimate authority to act? That's what Vendan Ananda Kumararajah has been digging into, and it's a really crucial point we can't afford to ignore. ### The AI Risk Map: A Good Start, But Not Enough Think of it like this: we've drawn a fantastic map of all the potential potholes and detours on the road of AI development. We've identified biases, privacy concerns, and even the possibility of AI making choices we don't understand. That's a huge step forward, and it's absolutely necessary for responsible innovation. We wouldn't want a self-driving car on the road if we hadn't mapped out every possible hazard, would we? However, just knowing where the risks are doesn't tell us when the AI is *fit* to drive. It doesn't tell us if it can handle unexpected situations, ethical dilemmas, or even just a tricky merge on the highway. That 'fitness to act' is a whole different ballgame. ### What Does 'Legitimate Authority' Even Mean for AI? When we talk about a human having legitimate authority, it usually means they've earned it through training, experience, and adherence to certain ethical standards. We trust a doctor with our health because they went to medical school, passed exams, and follow a strict code of conduct. But what's the equivalent for an AI? Is it just about crunching enough data? Or is there something more profound we need to consider? This isn't just a technical problem; it's a philosophical and societal one. We're essentially asking: can an algorithm ever truly *deserve* our trust in the same way we trust another human? ### The Human Element: Still Crucial Even as AI gets more sophisticated, the human element remains incredibly important. We can't just hand over the keys and walk away. There needs to be clear accountability. If an AI makes a bad call, who's responsible? The developer? The company that deployed it? The person who approved its use? This is why robust governance frameworks aren't just about preventing harm; they're about establishing clear lines of responsibility and ensuring that there's always a human in the loop, even if that loop is very wide. It's about maintaining oversight and the ability to intervene when necessary. ### Building Trust, Not Just Functionality Ultimately, the goal isn't just to build AI that *works*, but AI that we can *trust*. And trust isn't something you can program in with a few lines of code. It's built on transparency, accountability, and a deep understanding of an AI's limitations, not just its capabilities. Think about it: would you trust an AI to manage your retirement savings if you didn't understand how it made its decisions? Probably not. We need to move beyond just mapping risks and start defining the criteria for legitimate authority and fitness to act. Only then can we truly harness the power of AI responsibly and ensure it serves humanity's best interests. - Defining clear ethical boundaries for AI decision-making. - Establishing independent auditing processes for AI systems. - Developing standardized metrics for 'fitness to act' in various domains. - Ensuring human oversight and intervention capabilities are always present. - Creating legal frameworks that assign accountability for AI actions. This isn't an easy conversation, but it's one we absolutely have to have. Because the future of AI, and its impact on our lives, depends on it.