Fully autonomous robots are closer than you think- opinion

But before being impressed or scared, there is one important question worth asking: Where is the human?

A polished video rarely tells us how autonomous the machine actually is. Is somebody holding a controller outside the frame? Is an operator using VR goggles or motion-tracking gloves? Is a human supervising the robot remotely and intervening when something goes wrong? How many attempts happened before the successful one was published?

None of this makes the progress in robotics less impressive. It simply means we need to distinguish between a robot performing a task and a robot independently understanding, deciding, and completing that task.

The automotive industry already gave us a useful way to think about this.

At Level 0, the human does everything. As we move through the levels, the vehicle gradually takes more responsibility. At Level 4, it can operate independently within defined conditions. Level 5 is the destination, where no human driver is required. So how close are we to the Level 5 robot?

A robot performing successfully in a controlled environment is very different from a machine operating independently for hours in the real world. But there is another category of robots where this conversation becomes much more immediate.

A drone does not need to understand the entire human world. It may need to navigate from one point to another, recognize a specific object, avoid obstacles, coordinate with other systems, and continue operating when GPS or communications are degraded.

A relatively inexpensive machine can force a human being to hide, move, expose himself, or become a target.

Our long-term answer cannot simply be putting another soldier behind a screen and asking him to detect the threat and react faster. Counter-drone systems are becoming more autonomous too.

A defensive system can increasingly detect an incoming drone, classify it, track it, and direct a response with less human workload. That response might be another drone, electronic warfare, directed energy, or an automated weapon station.

The important change is not which countermeasure wins. It is that machine detection is beginning to meet machine response. This is where robot versus robot warfare starts becoming much less like science fiction.

Humans should remain responsible for the rules, authorizations, and consequences surrounding lethal force. But there is an important difference between keeping a human responsible for the mission and requiring that human to manually control every second of it.

That distinction may define the next generation of military autonomy.

There is another reason this transition is accelerating. Cost. If an attacking drone costs $500 and defeating it requires a $500,000 interceptor, the defense may be technically successful while economically losing the fight.

It is also about who can build enough capable systems at a cost they can sustain. Cheap enough to produce in large numbers. Simple enough to deploy. Replaceable when lost. Smart enough to coordinate with other systems. This changes the equation for smaller countries.

For most of history, population and industrial scale have been major components of military power. A larger country could mobilize more people, manufacture more equipment, and absorb greater losses. Autonomous systems will not eliminate those advantages, but they could weaken the relationship between population size and the amount of defensive mass a country can generate.

A small democracy cannot manufacture another million citizens. But it may be able to manufacture tens of thousands of machines.

An aggressor considering an invasion may therefore have to calculate not only the defending country's soldiers, aircraft and armored vehicles, but potentially thousands of inexpensive aerial, ground and maritime systems able to detect, coordinate and respond. The strategic value of autonomy may ultimately be greater than replacing individual soldiers.

It could change the calculation made before those soldiers are ever sent to war. For Israel and other smaller democracies facing significantly larger potential adversaries, that should be an important part of the autonomy conversation.

Are we ready for the Level 5 robot? Probably not.

At least not the general-purpose machine that can walk out of a laboratory tomorrow and independently understand our complicated physical world. But Level 5 may also be the wrong benchmark.

We do not need full autonomy everywhere before autonomy begins changing warfare.

A drone can achieve very high autonomy for one narrow mission. A counter-drone system can do the same for another. A ground robot can independently perform part of a logistics mission. Multiple systems can increasingly coordinate their actions.

Step by step, the role of the human changes from controlling every movement to supervising the mission. That is the transition I believe we should be watching. The most consequential autonomous robot of the next few years may not be the humanoid folding laundry in a viral video.

It may be a relatively inexpensive machine operating on a border or battlefield, facing another machine, while the human who once would have stood there remains somewhere safer.

That future is much closer than we think.