In a context of increasing tension (just yesterday the controversy over the "Chinese spy balloon" which would be flying over American nuclear sites) which sees the use of the technology becoming increasingly widespread, the Pentagon is exploring new options. Options that increase the capacity and effectiveness of military operations, especially those that involve advanced technologies such as drone swarms, artificial intelligence and robotics. The latest project, called AMASS (Autonomous Multi-Domain Adaptive Swarms-of-Swarms) is really worrying.
A “swarm of swarms”
The ambitious plan of the US Department of Defense, developed with the DARPA, contemplates autonomous and synchronized assaults through hordes of drones coming from everywhere: air, land and water. An idea that raises many concerns about man's ability to supervise these complex "swarm of swarm" operations. Unless you also want to leave autonomy on how to attack, destroy and kill targets.
In the documentation provided most of the details are classified. What is known, as mentioned, is that the system will allow the attack on multiple groups of aerial, terrestrial and underwater drones (equipped with a mix of weapons and instruments such as GPS, radar jammers and more). The other known detail is that AMASS will not require human assistance, as the swarms will be able to coordinate over an operational area such as an entire country. However, the Pentagon says, there will be people to supervise the system and ready to intervene if necessary.
Are we sure?
The tender (worth nearly $80 million) involves building a system to exchange information and coordinate the actions of thousands of units. Suppliers have until February 10 to submit an offer: the goal is to reach phase two, and carry out experiments with swarms of both real and virtual drones, gradually increasing their number and the difficulty of the missions. Until thousands of drones work together as a single destructive “cloud”.
Secondo Gregory Allen from the Center for Strategic and International Studies in Washington DC, the Pentagon has actually already tested swarms with hundreds of military drones, but adding land and marine components in larger swarms leads to increased complexity and can make communication a problem . “If drones operate in a region where communications are blocked, supervision becomes more difficult,” he says.
How much human supervision is possible?
This is a key question. To ask it is Zak Kallenborn from the Schar School of Policy and Government at George Mason University in Virginia. And I don't like the answer one bit. Also because I already know her. The conclusion is always the same. A topic I care about particularly, because I identify it as a "horror of the future" which we should try to escape, and instead I notice more and more signals in this direction.
The prospect that these drone swarms could use lethal force without direct human supervision is growing, it is clear. However, it is not clear how AMASS will address the issue. Although the US has recently updated its policy on autonomous weapons, it is unlikely that such a system will not be equipped with the ability to “harm” completely on its own.
And when thousands and thousands of drones are involved, the margin of error becomes narrow: nothing is enough. Hard times await us in this field.
