Eric White And that AI is what we are talking about today. And as somebody who has been bombarded with buzzword pitches over the past maybe four or five years or so, I have to say whoever wrote this particular one was kind of a breath of fresh air. Hearing that, you know, talking more about the limitations of this technology rather than just spouting off all this stuff that it’s not necessarily capable of doing yet. So why don’t we kind of just start where you stand currently on what you think all this is going and what are the autonomy standards that everyone is operating within even if they might be expanding upon them when they’re, you know, making those pitches to media entities and government contractors

Ben Wolff Sure, so first off, I think about the world of AI really being divided into two different buckets. One is the kind of AI that you and I might be using on a daily basis where we’re asking questions of this artificial intelligence set of algorithms. Those algorithms have access to massive amounts of data and we use those agents to be able to help us make better decisions as human beings. So it’s a knowledge-based type of AI. And that’s not what we do. That’s probably what 95% of the AI companies out there do, but that’s what we don’t do. What we do is we deal with algorithms that sit on machines that enable those machines to be able to interact with the real world. So think about the kind of AI that might be in a self-driving car or in an autonomous plane. That’s the kind physical world embodied AI that we do, and so your question was, where are we today with that? We currently have our AI flying on drones that allows the drone to identify what’s going on in its environment, to perceive through its sensors what’s on the battlefield, for example, and then to make adjustments in real time without requiring human direction or input to be able to evolve and change and adapt to what’s gone on on the battlefield. So we’re taking a machine that is otherwise remote controlled, meaning it’s being managed by a human operator, and we’re enabling that human operator to not have to focus on each and every adjustment or movement of the aircraft, and instead allowing a single drone to truly become fully autonomous, to be self-driving, if you will, based on the inputs from the environment. And then we take it a step further and we allow multiple drones from different manufacturers relying on our software to be able to communicate and collaborate with one another, so that all of the knowledge across all of drones is shared among the drones, and the drones as a swarm can then respond in real time to what’s happening.

Eric White So you’re talking about reading and reacting, as far as the reaction portion of it, what is the role then in humans interacting with it and reacting to those reactions? I mean, how do you determine how far you want to let it go on its own when it is making those necessary adjustments that you described.

Ben Wolff So our system is flexible enough that the humans can intervene whenever they want to. So what the human does is the human first defines the mission parameters upfront, determines which drones will be launched as part of a swarm, and then the human can observe the input and data coming off of every one of the drones and how they’re actually functioning. A really clear example that we build in is if we’re talking about a kinetic solution where the drone will actually, not just be observing but will actually interact on a kinetic basis, that ultimate decision to strike a target is always left to the human. But beyond that, we can define any number of parameters, requirements or conditions that require human engagement and interaction and, in effect, a supervisory role from the human that can be defined in advance when the mission is first planned and set.

Eric White Alright, so then million-dollar question. Is that true automation? I mean, how many levels of it can you actually get to when you’re saying, oh, it’s automation. It’s automated, but the human’s always making the decision. Achieving that true automations in probably most people’s minds is, you know, it is a crock pot. I set it and forget it. Is that ever going to be achievable for something that, you know, particularly on the tactical edge during a war? Is that something we want?

Ben Wolff So first of all, there’s a distinction between automation and autonomy. The way we think about that is automation is, it’s pre-programmed by the human, and the machine goes off and does it without variation. So it can’t respond in automation, it can respond to what’s going on in the environment because it’s a set it and forget it. You pre-program flight paths or waypoints, and you set it, and forget, and then that’s done. That’s true automation, that’s not what we do. We do autonomy. Where the machine itself makes decisions not based on human input, but it makes decisions based on what it’s understanding about the environment in which it’s operating. So that’s autonomy. We are true autonomy. So then you ask the question, is that something we want? Is true autonomy something that we want and where do you draw the line? From a technology perspective, we enable 100% true autonomy, it’s up to the human operator to decide when does the human want to put the brakes on. So our technology is fully autonomous. It’s just a question of what we as humans want to impose as breaks, or gates or hurdles before the machine pursues a tactical solution.

Eric White Gotcha. Okay. And so go looking towards the future. What is the capabilities of newer technologies coming on? And are we being sold some overestimation then when most people are talking about automation, but it’s almost as if they’re tricking, you know, consumers, investors, all those people with thinking that they’re selling autonomy, but there actually selling automation?

Ben Wolff I think that’s a very valid point. And the problem, of course, is we don’t have a lot of common standards that have definitional rigor. So yes, a lot of companies talk about autonomy, autonomous this, autonomous that. And in reality, they’re either talking about unmanned or they’re talking about automation, but they’re really not describing full autonomy. And that’s one of the things that I think it’s really important for everybody to understand. There’s remote controlled unmanned systems. Then those unmanned systems can either be automated, in which case the humans have made all of the decisions in advance and the machine can’t respond in real time, or then you get to full autonomy. And it’s kind of like the difference, we’re all familiar with self-driving cars, Tesla’s done an amazing job of showing what the future looks like with true full autonomy, but right now they’re still not there. You can’t go on a back country road and just set it and forget it and have it respond to what’s going on in the real world environment. And I think we’re seeing the same thing with weapons systems. You’ve got this evolution of increasingly autonomous systems. In our case, we are true level five autonomous solutions for flying machines.

Eric White Looking ahead is the feasibility and affordability and all of those things that are factoring in, are those going to be less factors, I guess, would be the question where it’s not really, you know, you’re kind of weighing,  this costs a lot, but we’re getting autonomy as a reward for paying all this money and putting all this manpower towards it — is it going to worth it in the end? I guess is the question I asked, or it will we end up paying for a very expensive system that we need to revert to human expertise for anyway in the end?

Ben Wolff I think where we’re seeing the technology go is this technology is going to be more capable than what any group of humans could do on their own and that it will be more reliable and more precise because you take out human judgment and error in some of the more mundane tasks. So when you hear about the Department of War talking about wanting to be able to launch a thousand drones simultaneously, with today’s technology, before our system came along, for the most part, you’re talking about either a thousand soldiers having to fly a thousand drones. That’s not scalable from a manpower perspective. Or, you’re talking about launching a thousand drone that are entirely pre-programmed, which in a lot of use cases, I mean, that means the offense isn’t taking into account in real time what the defense is doing, right? And, you know, I think a lot about using a football analogy when the head coach calls in a play to the quarterback and the offensive line is ready to go, that’s great if the defense does exactly what you think they’re going to do. But as soon as they throw you a curve ball — sorry, I’m mixing my baseball and football analogies — but as soon they give you a different defensive look, you know, right before the ball is hiked, now the offensive line has to use their own brains to evolve and adapt to what they’re seeing the defensive line do. And that’s the same thing you see on a battlefield. So yeah, all the pre-planning in the world is great until the action starts and the plan falls apart. And that is when you need real time intelligence on the battlefield and that’s what we do.

Eric White And as far as securing the autonomous system itself, I mean, to continue the football analogy, you don’t want anybody else having to look at the playbook. How is that going to be taken care of and is the answer to that even more autonomy? So autonomous protection, protecting the autonomous drones to the point where we are so many levels that we’re getting into strange territory?

Ben Wolff Well, it is definitely strange territory, but we have to be prepared from a cyber perspective to be able to protect the systems that we’re deploying. And that’s true. Anything that’s got communications in road into it, we’ve got to be to secure and protect. And that is fundamental for every weapon system that’s out there. That’s true for manned aircraft as well as unmanned aircraft. So that is a real world problem that everything has to deal with. Now, one of the things that we bring to that particular discussion is our AI lives on the edge on the machine, not in some central server somewhere that can be hacked into. So our whole system, one of the things that’s unique about it is the intelligence lives on each individual drone as part of a mesh network created by multiple nodes on the network. You don’t have to consistently go back and have continuous communications to a central server, somewhere. That helps enhance the overall security and safety of the system that we deploy.