A Morning Inside the Mind of a Futurist — Where Every Question Leads to a Drone

A stream-of-consciousness journey through the AI-controlled skies of our immediate future

By Futurist Thomas Frey

Every morning begins the same way — in that warm, blurry corridor between sleep and awareness where the best thinking happens and no alarm clock is welcome.

The transition from dream-state to conscious thought is never clean. Ideas arrive half-formed, like signals from a frequency that only tunes in clearly for a few minutes before the day crowds them out. This morning, before my feet touch the floor, a single question presents itself with unusual clarity:

Who — or what — is actually in control of the drones?

That question alone will occupy the next hour of my thinking. Maybe longer.

The AI Brain No One Is Talking About Enough

We talk constantly about drones. Drone delivery. Drone warfare. Drone photography. We talk about them as if they are the story.

But drones are not the story. The AI controlling them is.

Think about what is actually happening right now, in 2026. The Pentagon has launched a $100 million prize competition specifically to develop voice-controlled, autonomous drone-swarming systems — technology that translates a spoken sentence into coordinated action by dozens of drones simultaneously. SpaceX and xAI are reportedly competing. So are defense giants. The goal is explicit: a single operator managing a distributed swarm of autonomous systems, each running its own onboard intelligence, communicating laterally with the others, making real-time tactical decisions without waiting for a human command.

That is not a drone program. That is an AI program that happens to have wings.

And the cascading questions begin.

Input: How Does an AI Drone Actually Receive the World?

If the AI is the brain, what are the senses?

Today’s AI-controlled drones perceive their environment through a layered stack of inputs that would have seemed like science fiction a decade ago. LiDAR maps the physical space in three dimensions at hundreds of thousands of points per second. Computer vision identifies objects, tracks movement, and classifies threats or targets in real time. Acoustic sensors detect audio signatures. Environmental sensors read temperature, humidity, and air pressure to self-correct flight paths. GPS data is cross-referenced with onboard inertial systems so that the drone keeps flying accurately even when GPS is jammed or degraded.

Then the AI does something remarkable: it fuses all of those inputs simultaneously, builds a probabilistic model of the world it’s operating in, and acts — in milliseconds.

So here is my question: at what point does the quality of a drone’s sensory input exceed the quality of a human pilot’s? At what point — if it hasn’t already happened — does the AI see better, hear better, and react faster than any human ever could?

I think we passed that threshold somewhere quietly in 2025, and most people missed it.

When autonomous swarms make collective decisions, the technology advances faster than our ability to define responsibility, accountability, and control.

The Swarm as a Single Distributed Mind

Now take one AI-controlled drone and multiply it.

The U.S. Army’s autonomous swarm program uses hierarchical reinforcement learning — meaning the swarm learns from every mission, every failure, every environmental variable, and self-improves over time. Ukraine’s battlefield deployments have already demonstrated coordinated operations of 3 to 25 drones per mission in over 100 documented combat engagements through 2025. The Perdix micro-drone system demonstrated something even more profound: a “distributed brain” architecture in which no single drone is in charge. If you take out half the swarm, the remaining units self-reorganize and continue the mission.

That is not a tool. That is a new form of collective intelligence.

Which leads to a question I cannot stop turning over: what does it mean for AI to make a collective decision? When a swarm of 50 autonomous drones reaches a fork in the mission — abort or proceed — and there is no human in the loop, who is responsible for that choice? The software architect who wrote the objective function? The commander who authorized the mission? The company that trained the model?

We do not have good answers. We are flying faster than our ethics can follow.

Output: What Does an AI Drone Actually Do in the World?

The military applications are the ones that dominate the headlines, but they represent only one end of a very long spectrum.

On the civilian side, the output of AI-controlled drones is already reshaping daily life in ways most people have not registered. Zipline — the company that pioneered autonomous medical drone delivery — has now completed over two million commercial drops worldwide. Advocate Health is planning what would be America’s largest drone-based pharmacy operation, targeting 100,000 annual flights and delivery times of five to seven minutes from autonomous launch docks to doorstep. The global medical drone delivery market, worth $166 million in 2025, is projected to reach $2.1 billion by 2035.

Think about what that output means in human terms. A patient in a rural county has a cardiac event. An AI dispatches a drone carrying an automated defibrillator before the 911 dispatcher has finished taking the call. The drone arrives in four minutes. The ambulance arrives in eighteen.

That is not a technology story. That is a life, saved or not saved, depending on whether a swarm was flying or sitting in a hangar.

The moment we design systems to operate beyond our reach, we must ask whether autonomy is expanding capability—or quietly reducing control.

The Questions That Keep Me in Bed Longer

Here is where my brain refuses to stop:

If AI controls both the input and the output — if it is sensing the world independently and acting on it autonomously — what exactly is the human’s role? Supervisor? Author of goals? Ethical override button?

And if the AI makes a mistake — if a misclassified object leads to a wrongful action, whether in a delivery gone wrong or something far more consequential — how do we even reconstruct what happened inside the model’s decision process? The drone knows. The log file might know. But the reasoning that led from input to output may be as opaque as the inside of a black box dropped in deep water.

By 2026, additional emphasis has been placed on AI drone systems that can operate in communications-degraded or denied environments — meaning the drones are designed to function when no human can reach them even if they want to. That is a profound architectural choice with profound implications. We are building systems intentionally designed to operate beyond our reach.

Does that mean we are building systems beyond our control?

The Question Underneath All the Questions

I think the most important AI drone question is not technical. It is not about sensor fusion or swarm coordination or edge computing latency.

It is this: whose values does the AI encode?

Every autonomous system is built on an objective function — a mathematical definition of what the AI is trying to achieve. Maximize delivery speed. Minimize civilian exposure. Neutralize the target. Those objectives are written by humans, and they carry every assumption, bias, and blind spot of the people who wrote them.

The commercial drone market reached $38.2 billion globally in 2025 and is projected to reach nearly $190 billion by 2034. That is an enormous amount of autonomous behavior being encoded into the world. Each autonomous decision a drone makes reflects a value judgment made by someone, somewhere, at some point in the design process.

Which means that designing AI-controlled drones is not a technical problem. It is a moral one. And we are not treating it that way.

The future is not defined by the technologies we can build, but by whether our wisdom evolves as quickly as our capabilities.

Finding Focus in the Questions Themselves

I’ve been asked many times how I think about the future. What tools I use. How I bridge the gap between where we are and where we’re going.

The honest answer is that I follow the questions. Not to answers — answers have a way of closing doors — but through them, like a corridor where each door you open reveals three more.

AI-controlled drones are not a destination. They are a corridor. And we are walking it right now, faster than most of us realize, with systems that sense, decide, and act on our behalf in ways we are only beginning to understand.

The real question is not whether we can build them.

It is whether we are wise enough to deserve them.

Interesting Engineering — SpaceX Joins Classified Pentagon Bid for Voice-Controlled Drone Swarms https://interestingengineering.com/military/spacex-drone-swarms-for-pentagon

DefenseScoop — Pentagon Preparing for Drone Swarm ‘Crucible’ — AI Inter-Agent Collaboration https://defensescoop.com/2026/03/31/pentagon-preparing-drone-swarm-crucible/

AI CERTs News — AI Drone Delivery: Transforming Emergency Healthcare Logistics https://www.aicerts.ai/news/ai-drone-delivery-transforming-emergency-healthcare-logistics/