The article gives a good overview of the situation. I have only a few things to add.

• war is a match of game theory (finding moves that harm the opponent) and accelerated evolution of the technology and methods for doing so

• if a weapon works, choosing not to use it typically increases chance of defeat

• there is no doubt that autonomous weapons work

• about the only thing which seems to prevent some weapons (e.g. of mass destruction) from being used in some wars, is threat of being targeted with the same kind, or a large coalition of opponents forming and intervening

• so, only continued development of international law (collective agreement on behalf of a large coalition of states, representing most of human population) has any chance of restraining use of autonomous weapons

• sadly, I don’t see states wanting continued development of international law

• restraining production of autonomous weapons is probably hopeless, a reasonably good competition robot can be weaponized with some success

So, future will have states threatening each other with swarms of autonomous weapons, and terrorist entitities using them at a smaller scale. Weapons that destroy electronics (EMP, microwave, lasers for frying cameras) will be highly sought after for defense. Air defense and antidrone systems will be much more common than today. At some point in history, there will be a war where a large swarm of autonomous weapons is used for mass destruction (maybe against civilians too). Quite possibly, a response will be given with a different weapon of mass destruction (e.g. nuclear EMP used to disable the swarm). Regulation is unlikely to succceed since states are nowhere near agreement (in fact, fighting new and avoidable wars among each other).

Undoubtedly, the most advanced systems in the world are being devised in laboratories in the US and China. A Pentagon programme known as Replicator 1 is due to deliver “multiple thousands” of all-domain autonomous systems by August 2025.

Why is this undoubtedly? Fact is, most other countries right now are lagging behind in this field and without extensive field testing whatever these labs will produce will be years behind state of the art. You can’t design these things blind and both US and China actually lack people that have experience in drone warfare or it’s years outdated so design decisions beyound basics will be worse, then those that were made with consultations and input from people that actually using the drones. There is no doubt that both US and China can pour a lot of money into it, and make great machines from engineering standpoint and they won’t perform as expected. Drone warfare in current form is pretty new and ideas iterate very fast, and army procurement is anything but fast.

Cyberdyne Systems approves

AI taking notes…