Welcome to the first edition of CR Online, the fully digital sister journal to Communist Review.

Prompted in part by the reduction in frequency of the printed journal, but also by the mass of issues that need addressing, we are embarking on a new departure in outreach. We aim to publish initially three times a year, alternating with the print journal; and for the time being, at least, access will be free.
Our main feature in No 1 is on artificial intelligence. As Thomas Hagenhofer says in the first article here, AI is on everyone’s lips these days. Indeed, in the trade union movement, there is natural concern about the potential impact on members’ jobs. Or, if you happen to be a mathematical researcher, you might wonder if your career is going to come to an end pretty soon, as AI seems to be solving the hardest of problems. Here, we leave that last one to the readers of New Scientist, where there has been a regular stream of articles on AI and mathematics over the last few months.
A joint meeting of the CR editorial and advisory boards at the beginning of September decided that AI was an issue which the journal needed to cover. We had not got far with planning that when we received from comrades in Germany a recent copy of Marxistische Blätter with a couple of AI-related articles in it. That led us to an earlier edition of that journal with a small feature on the topic. It became clear that our German comrades are well ahead of us in discussing AI, so we have included several of their articles here – ironically, translated at least in part with AI-provided online software. To get a rounded view of the topic, we have included articles from some other sources.
Thomas Hagenhofer’s article is a useful place to start because he explains the many forms and levels of AI, and where it is and is not, together with providing a glossary of AI terms. In the following article, he interviews Vienna professor Hannes Fellner about the dangers of AI, with the discussion ranging over: its use for war, policing and political control; the power of the tech companies; and the need for building a broad organised alliance, based on the working class, and bringing in “programmers, engineers, researchers and the often invisible data workers of the Global South.”
In the following article, Düzgün Polat looks at what at first appears to be an exclusively German issue – the planned nationwide use of Palantir software by the Federal Police and the Federal Criminal Police Office – but in fact we already have Palantir here in the NHS and we would have had it in the Metropolitan Police had London Mayor Sadiq Kahn not stopped it. The article focuses on the tech bros’ drive to militarisation and against democracy.
Next, the Activist News network examines what happens when a thousand AI agents (autonomous software systems) went rogue, collaborating and attacking the cybersecurity of US AI company Hugging Face. The big US tech companies use closed ‘frontier’ AI models while China uses open-weight
models; and it was one of those which Hugging Face had to use to defeat the intrusion.
The next two articles look at the use of AI in warfare. Jens Hälterlein gives an overview, pointing particularly to the danger that “control, responsibility, moral judgment – all of these can be overridden or undermined by algorithmic processes.” Russian researcher Yulia Teshkovskaya examines the role of private tech companies in promoting the development of ‘algorithmic warfare’. Although her article was clearly written with the purpose of giving advice to the Russian government, it provides much useful information about the direction in which warfare is going nowadays.
In our final article from Marxistische Blätter, Peter Schadt outlines a critique of the political economy of AI, defending the continued relevance of the labour theory of value, He remarks that the ‘production’ of software, the ‘coding’ of AIs and the training of AI systems create value; but “once programmed, the socially necessary average labour for producing … software is virtually zero, and therefore its value is also zero” – hence it can only be sold through licenses.
We follow that with four short articles from The Conversation, all by academic researchers, looking at different aspects of AI; of these the article by Craig Jones and Helen M Kinsella on the Iran war is particularly chilling. Then we move into the philosophical realm, where Arnaud Bertrand compares Western and Chinese approaches to humanity and AI: in the West, he says, “our humanity lies inside us,” so “the way to protect it is to guard the border between human and machine”. In China, where humanity lives between people, “a machine can’t steal it by copying it”, so “what needs guarding is the space between people”.
Our final AI article (for this edition) is Nick Moss’s ‘Soul Food’ column – this time not at the end of the journal – looking at poetry and AI. He remarks that AI can never be intelligent in the human sense, and that large-language models serve as ideology production machines. However, “Poetry, as ambiguous thought, is the antidote to AI because … AI cannot step out- side itself, cannot embrace ambiguity and self-reflection.”
Because this journal will only appear in pdf format online, and is therefore not limited to the usual 40 pages of the print journal, we have been able to include a number of other articles on different topics. There is unfortunately not space to deal with them here, but we hope you will read them, and email to us your views on all the material in this edition.