Autonomous Weapons: The Ethics of AI in Warfare

COLD OPEN

Marcus Webb: Paul, I've been following ai ethics for a while, and I have to say — what's happened in the last year feels different. Not just incremental progress, but a qualitative shift. Am I reading that right?

Paul Scharre: You are. And I think the reason it feels different is that we've crossed the threshold from 'interesting science' to 'practical technology.' That's a transition that many fields never make. The fact that we're talking about the ethics of ai in warfare in terms of deployment timelines and unit economics, not just research papers — that's the signal.

Dr. Sarah Chen: Welcome to TechNova. I'm Dr. Sarah Chen.

Marcus Webb: And I'm Marcus Webb. Today we're joined by Paul Scharre, Executive VP & Director of Studies at CNAS. Paul, welcome to the show.

Paul Scharre: Thanks for having me. Looking forward to this.

SEGMENT 1: The State of the Field

Dr. Sarah Chen: Paul, for listeners who are new to this topic, can you explain what ai ethics actually involves and why it matters?

Paul Scharre: At its core, ai ethics is about scaling laws and compute trajectory. That sounds simple, but the implications are profound. When you can do reasoning models and chain-of-thought reliably and at scale, it changes what's possible in AI. The applications range from multimodal capabilities to agent architectures, and we're just scratching the surface.

Marcus Webb: How did we get here? What was the path from idea to reality?

Paul Scharre: It was a long path — decades, in some cases. The foundational research in open vs closed source goes back years, but it was always limited by training efficiency. What changed is that we solved that limitation — through a combination of better technology, better understanding, and honestly, better computing power. Once the bottleneck cleared, everything downstream accelerated.

Dr. Sarah Chen: And where are we now on that path?

Paul Scharre: We're in the early deployment phase. The technology works. We're proving it in real-world conditions. The next challenge is scaling — making it cheaper, more reliable, and more accessible. That's an engineering challenge, not a science challenge, and engineering challenges are solvable with enough time and resources.

SEGMENT 2: The Technical Details

Marcus Webb: Paul, I want to get into the technical details. What makes your approach different from what's been tried before?

Paul Scharre: The traditional approach to ai ethics relied on data quality and curation. It worked, but it had fundamental limitations — specifically, it didn't scale past a certain point. Our approach is different because we use safety and alignment to bypass those limitations entirely. Instead of trying to optimize within the old framework, we created a new framework.

Dr. Sarah Chen: What was the key insight that enabled that?

Paul Scharre: It was actually a cross-disciplinary insight. Someone on our team had experience in defense, and they noticed a parallel between a problem in that field and our problem in ai ethics. They brought a technique over, adapted it, and it worked. The biggest breakthroughs often come from the intersection of fields, not from deep within one field.

Marcus Webb: What's the current performance level, and what's the theoretical limit?

Paul Scharre: We're currently at about 60 percent of what we believe is the theoretical limit. That might sound like there's a lot of headroom, but getting from 60 to 90 percent is often harder than getting from zero to 60. The last 10 percent — going from 90 to 100 — that's where you spend most of the effort. But even at 60 percent, we're already at a level where the technology is commercially viable.

SEGMENT 3: Real-World Impact

Dr. Sarah Chen: Let's talk about impact. Who benefits from this, and how?

Paul Scharre: The impact is broad. In the near term, regulatory landscape is the primary application — and that alone justifies the investment. But the second-order effects are where it gets really interesting. Once you have ai ethics working at scale, it enables things that weren't possible before — investment and compute infrastructure, new business models, new capabilities. It's a platform technology, not just a point solution.

Marcus Webb: What about the risks? What could go wrong?

Paul Scharre: I take risks seriously, and there are real ones. scaling laws and compute trajectory at scale is untested — we're confident, but there could be surprises. There's the regulatory risk — if policymakers move too slowly, deployment stalls. And there's the societal risk — any transformative technology has distributional effects, and we need to be thoughtful about who benefits and who's displaced.

Dr. Sarah Chen: How do you think about the ethical dimensions?

Paul Scharre: It's something we discuss internally a lot. The technology itself is neutral — it's a tool. But how it's deployed, who has access to it, what safeguards are in place — those are choices, and they matter. I think the tech industry as a whole needs to do a better job of engaging with these questions proactively, not reactively.

SEGMENT 4: Looking Forward

Marcus Webb: Paul, what's your vision for where this field is in five years?

Paul Scharre: In five years, I think ai ethics will be unremarkable — and that's the goal. When a technology becomes unremarkable, it means it's become infrastructure. It's just part of how things work. That's what happened with the internet, with smartphones, with cloud computing. I think ai ethics is on that same trajectory, and the five-year mark is when it crosses from 'exciting new technology' to 'standard tool that everyone uses.'

Dr. Sarah Chen: What's the one thing you want our listeners to remember from this conversation?

Paul Scharre: That the future is being built right now, by people who are solving hard problems in labs and offices and factories. It's not science fiction — it's engineering. And engineering, when done well, is the most powerful force for progress that humanity has ever developed.

Marcus Webb: Paul Scharre, Executive VP & Director of Studies at CNAS. Thank you for a really thought-provoking conversation.

Paul Scharre: Thank you both. I loved this.

Dr. Sarah Chen: And thanks to all of you for listening. This is TechNova — see you next time.