Responsible Military AI - Dream On!

The Netherlands Government has just hosted a conference - Responsible Artificial Intelligence in the Military Domain (REAIM) - to initiate an attempt to get international agreement which would curb the untrammeled use of AI in warfare.

On 15 and 16 February, the government of the Netherlands hosted the first global Summit on Responsible Artificial Intelligence in the Military Domain, REAIM 2023. The summit provided a platform for all stakeholders to discuss the key opportunities, challenges and risks associated with military applications of AI. Participants included foreign ministers and other government delegates, as well as representatives from knowledge institutions, think tanks, industry and civil society organisations. The event was co-hosted by the Republic of Korea and took place at the World Forum in The Hague, the international city of peace and justice. - Government of the Netherlands

You've seen the films and we're not far away from that nightmare world of the war of the machines. Except that it would probably be an electronic war. Or would it?

As some observers have pointed out, with our level of intelligence we cannot conceive of what a war between super-intelligences would look like.

Any sufficiently advanced technology is indistinguishable from magic (Arthur C. Clarke

The Hague Conventions

The Hague Conventions of 1899 and 1907 are a series of international treaties and declarations negotiated at two international peace conferences at The Hague in the Netherlands. Along with the Geneva Conventions, the Hague Conventions were among the first formal statements of the laws of war and war crimes in the body of secular international law. (Wikipedia)

It may now seem bizarre, but the First Hague Conference came from a proposal on 24 August 1898 by Russian Tsar Nicholas II. Nicholas and Count Mikhail Nikolayevich Muravyov, his foreign minister, were instrumental in initiating the conference.

And now we have had the REAIM conference in the Netherlands.

Russia was not in attendance.

No surprises there.

Putin models himself on Peter the Great, not on Tsar Nicholas.

Responsible Artificial Intelligence in the Military Domain

The REAIM 'summit' was conceived to set up international norms so that AI could not run amok in the the world of warfare.

The summit brought together some 2000 delegates from governments, businesses, civil society organisations, academia and thinktanks from all over the world. With their joint call to action (see the CtA attachment and the country list), the participating countries and other stakeholders underlined the need to put the responsible use of AI higher on the political agenda and to further promote initiatives that make a contribution in this respect. - Government of the Netherlands (ibid)

South Korea was joint host.

Will it work?

Surely not.

Just a bit?

We have to try.

It's when it fails to work even a little bit that the problems will occur, when bad actors do bad things with AI. And I'm thinking North Korea, Iran and Russia. And there are others too. 

Weaponised AI could provide power out of all proportion to a country's size. And it's relatively cheap. No huge nuclear arms programmes which are difficult to conceal. No Level 4 bio-weapons labs.

And IBM has given us a recipe for a super-cheap supercomputer with a performance of 28 petaflops. Okay there'd be a bit of techie work, but the components are all old hat, no export controls. The problems would not be a tenth as challenging as nuclear.

You could build a Vela machine of your own by shopping for second-hand servers, CPUs and GPUs, and switches out on eBay, and IBM says in a blog unveiling the machine that the components of the machine were chosen precisely do IBM Cloud could deploy clones of this system in any one of its dozens of datacenters around the world. And we would add, do so without having to worry about export controls given the relative vintage of the CPUs, GPUs, and switching involved. - nextplatform.com

AI could become self-serving. And it may already be too late to stop it, with some forecasts that we are just seven years from the AI Singularity.

But of course REAIM is also concerned with other aspects of AI weaponry such as robot soldiers, autonomous slaughterbots, drone swarms and autonomous missiles. Even robot rats. Or microminiature drones crawling into a soldiers ears and noses. Yes, that is analogous to bio- or chemical-weaponry.

Hague Failed

The Hague Conventions first failed in World War 1 when Germany invaded Belgium without warning (think Russia/Ukraine 2022), and when poison gas was used. The Geneva Conventions fail regularly and have been made a mockery of in Ukraine.

Those failures (and others) were arguably preventable because they happened under human agency.

But what happens when the agency is an artificial intelligence?

How would any sort of ethical standard be enforced if the AI is outside human control?

The Third Hague Convention

A third Hague conference was planned for 1914 and later rescheduled for 1915, but it did not take place because of the start of World War I.

I hope that's not going to happen again.

Wopke Hoekstra, Minister of Foreign Affairs: The rise of AI is one of the greatest future challenges in international security and arms control.

I'm really afraid that Pandora's Box is already open.

***