CONTRA EX MACHINA: AI & OTHER VIRTUAL PERILS by Verona Silenda (extract from NFB magazine issue 12)
One of the most important issues of our time is the growing influence, tantamount to colonisation, that the virtual/digital has on our lives. This article looks at that, especially AI.
NB: References below are all in Red. This article is from pages 34-37
MY CREDO
I hail from Left field, with a dash of anarchism feminism and Green politics. And, like all of us, am unimpressed by most (not all) of the Last Century Left (LCL). About this subject (with honourable exceptions) the LCL says little, other than cheerlead: which doesn’t count. ‘Fully Automated Luxury Communism’—you’re having a laugh, on so many levels. So, hitherto much discussion has been left to two groups.
Firstly, technophile insiders, such as Clay Shirky 1 and Ray Kurzweil 2, who want the onward digital/AI march to speed up even faster. Shirky thinks we will all benefit from being subsumed in the mass, and have loads of free time. Guess the only working class he meets are his chauffeur and nanny. Probably thinks Precariat is a new after-shave. Idiot. Kurzweil longs for the day we reach the ‘Singularity’ and our human bodies fully merge with machines. Count me out: have you seen the Borg lately? Ain’t pretty. And seriously, where is humour (and the oxygen of sarcasm) in this Timid New World?
Second group are interested insiders: not political, but worried, like Nicholas Carr: who has shown how the internet dumbs our brains 3, and the onward march of automation de-skills us bringing new perils 4. Then the amazing Jaron Lanier (critic of the ‘hive mind’, exemplified by that Citadel of Fools known as Wikipedia 5), who excoriates the pervasive inanity of social media 6. Also, intelligent observer James Barrat has interviewed, and intellectually deconstructed, many movers and shakers in this field 7. Unchecked, a bleak future lies ahead or maybe no future at all. Sitting uncomfortably? Let us begin….
EVERYTHING CONNECTING IS A PROBLEM
Connecting everything: house-hold appliances, information storage systems, financial applications etc. is presented as a panacea, ending having to wait for goods/services or, god forbid, not getting same/next day Prime delivery. ‘Just in time’ ordering systems in supermarkets and factories are great for the bottom line/reduced storage costs but a looming disaster in the real world of supply chains. Interconnectedness is a deep flaw: any ‘open’ nexus is vulnerable to hackers, as when NHS computer systems were corrupted in 2017 by the Wannacry virus: cost £92 million/19,000 appointments cancelled 8. Many NHS areas just cannot afford IT system upgrades. The interconnectivity of such systems (and abolishing paper records) makes any breach/glitch catastrophic. There is no compensatory resilience built in.
Unquestionably, large corporations mug the public, via outrageous lucrative contracts, and denying an IT system could ever be wrong. Hubris exhibited by Fujitsu’s Horizon System that criminalised many UK postmasters, leading to untold misery, backed by the courts 9, especially since April 2000, from which date computerised ‘evidence’ was accepted without question. Elon Musk’s ambition for X to become an ‘everything app’—payments, messages, video-calls etc fills me with dread 10. The civil liberties implications for digitally cancelled individuals are too great. The ‘Stuxnet’ virus by which Mossad/the CIA crippled Iran’s nuclear programme long ago escaped to cause havoc world-wide.
All infrastructural systems are vulnerable: murdered GCHQ/MI6 spy Gareth Williams was possibly scoping this when he died. A classic account of systemic vulnerabilities is the Joseph Weiss book ‘Protecting Industrial Control Systems From Electronic Threats’ 11. It will keep you awake at night even if your ‘Smart Home’ system hasn’t been hijacked with your devices all on full blast…Slowing things down, introducing physical firewalls, having where possible paper records as back-up, opposing digital currency: I favour all these.
PERSONAL DETAILS BEING DIGITISED IS A BIG PROBLEM
At the press of a button personal details of 10,000 Police Service of Northern Ireland staff were published on the internet in August 2023. Predictably, our old friend ‘Human Error’ was trotted out as explanation, along with sibling ‘Simple Human Error’ 12. Something like this has happened before, when the (RUC: as was) Castlereagh Detention Centre was raided in March 2002 and files containing details of informants were removed. This had MI5’s fingerprints all over it, and strange to say their asset inside the IRA (Denis Donaldson) was soon unavailable to answer questions (having been shot dead) 13. That raid took some organising, and staff present could hear the lead hoodlum had a very English accent. Far easier to acquire information digitally do you not think? In September 2023, with little publicity, 8,000 Greater Manchester Police details were purloined during a targeted attack on a sub-contractor 14. No error, but again, if these details had been kept ‘in house’ this probably wouldn’t have happened (although ex-Manchester cop Pete Jackson might disagree). If you detect a theme—systemic data liquidity as the problem, you’re right.
Wikileaks and Edward Snowden were both on the side of the angels in uncovering naughty state actions. But an issue arises from their work, brought sharply into focus by the April 2023 case when 21-year-old US National Guard member, Jack Teixera, accessed, downloaded then uploaded (onto the Discord messaging platform) classified Pentagon documents revealing, among other things, 50 of 97 overseas Special Forces in the Ukraine were British 15. Good to know, but is it not highly disturbing British secrets are so easily accessed overseas, by almost anybody? Not Five Eyes but 500,000 Eyes. While some of this ‘information sharing’ is a reaction to too little of that happening before the 9/11/01 attacks in the US, enough is surely enough. And no, we don’t think 9/11 was an ‘inside job’—check it out 16.
FAKE NEWS/DEEP FAKES
The biggest ‘fake news’ stories have been propagated by Western governments: using 9/11 as an excuse to invade Afghanistan, or the false claim Saddam Hussein was developing ‘Weapons of Mass Destruction’, which led to the Gulf War and 250,000 dead, yet no war crime prosecution of Blair/Bush. It isn’t just governments: Observer journalist Carole Cadwalladr hyped beyond all recognition the self-publicity of public school con artists Cambridge Analytica to suit a fictitious narrative that older (mostly non-web savvy) people who voted for Brexit in 2016 did so because Russian bots flooded the internet with propaganda. She included Brexiteer Aaron Banks in this fantasy, and was financially stung in the consequent libel action. Showing (contrary to that Orwellian-sounding censored Guardian forum, Comment Isn’t Free). There was an amusing spin-off: Series 2 of BBC’s ‘The Capture’ (2022) elaborated on this hokum for a hilariously comedic plot-line.
The September 2023 Slovakian Election may have been influenced by a ‘Deep Fake’ audio of the losing candidate planning to rig the election 17. Ken McCallum, Director General of MI5, said shortly after that such technology “could plant all kinds of confusion and dissension and chaos in our societies…a range of actors of various sorts…might seek to use these technologies to influence public opinion in all kinds of ways” 18. MI5 (like Russia’s FSB) are one of those actors, so we have all been warned.
THREE AI FACETS DEFINED
A common misconception is to see AI (Artificial Intelligence) as just a tool. Wrong. There are three AI developmental ‘stages’:
AI as it currently is, with some useful aspects: search engines are one—albeit censored—a worrying proof is Google’s Gemini image-generation tool, roll-out halted temporarily in February 2024 when it became evident white people were erased and historical figures events and buildings falsely portrayed 19. Whatever tweak is made, they are clearly in the propaganda business. Then there are Siri and Alexa. The latter is undoubtedly good for blind people, though always gets foxed when I ask if she works for the CIA. Try it! On the down-side, as of 23/3/24 there have been 500 deaths in Tesla cars, and 42 where Tesla Auto-Pilot was involved 20—raising legal and philosophical questions about responsibility/culpability.
AGI: meaning Advanced General Intelligence, AI that will have reached the level of a human: a good definition is Barrat’s “the ability to solve problems, learn and take effective human-like action, in a variety of environments” 21. You might say, what’s wrong with that? But at that point, having developed consciousness, the question will need to be posed: will such an intelligent machine be servant, colleague or competitor? A question already answered in Bill Joy’s April 2000 Wired article, ‘Why The Future Doesn’t Need us’. Other responses usually miss the point—the intelligence of AI beings, and speed of operation, would make human-speed counter action advised such as—“conceivably upon encountering a situation which goes too far, potentially threatening our existence or relevance we could intervene” 22 inadequate. Like a proponent of gun control saying if confronted by an armed assailant they might radically rethink their stance. Piffle. There are currently limits to AGI based on how machines learn compared to humans, but advances in processing speeds will soon overcome those 23. Kurzweil thinks AGI may be here by 2029. At this juncture, every reason to think development will speed up, exponentially, leading to…
Advanced Super Intelligent AI: will not necessarily be hostile to us as such, but why, exactly, would such a being need, or even empathise with us? Might it not, logically, come to see humanity as a virus, like in the wonderful (and disturbing) Matrix films? Egotistical (male) geeks run tech companies: Elon Musk, Bill Gates, Mark Zuckerberg, and the loony-tunes control freak Sam Altman of Open AI. None have a shred of empathy, or know what ethics is. In which case the resultant products, unlike say new medicine, will be (are being) dumped on an unsuspecting public, guinea pigs in this worldwide digital lab with no safety protocols. Kurzweil thinks ASI may happen by 2045: no certainty, but might it not be sagacious to consider the possibility and plan accordingly?
Many think conceiving of a negative AI outcome is a misnomer, confusing intelligence with consciousness. But if the two merge, even slightly, in ASI, the game for humans may be over. One route by which this might happen is the development of ‘Intelligence Augmentation’, whereby microchips inserted into the brain artificially enhance human intelligence. No science fiction—the Elon Musk-founded company Neuralink has already embedded a chip in a subject’s brain. Musk intends to create a mass-market device that would connect user’s minds with computers to achieve “symbiosis with artificial intelligence” 24—i.e. the Singularity. For PR reasons, trial subjects are currently people with significant spinal injuries, but the direction of travel is clear. Yet this was no properly regulated clinical trial, a first ‘success’ was announced by Musk in a Tweet. Going forward a fusion between artificial intelligence and consciousness, exactly what those who regard the likes of myself as misplaced Cassandras for predicting, is explicitly what Musk is now developing, unfettered by any regulations whatsoever. If that doesn’t worry you, your synapses must have fused…
AI IMPACTS
AI’s ongoing impact can be divided into three overlapping areas: the economic, technological/social and the security implications.
Economically, even (!) the International Monetary Fund is worried: which means we should be, that c.30% of jobs in ‘advanced’ economies will be lost, thus exacerbating already existent trends marginalising many of the traditional working class 25. Findings ostensibly echoed at the January 2024 World Economic Forum Davos summit, where—allegedly–“the threat of artificial intelligence stoking greater inequality and sparking a financial crash provided a stark background to discussions” 26. I say ‘allegedly’ because behind the scenes this—displacement of humans seen (de facto) as ‘useless eaters’ in that chilling phrase, by ever more powerful corporations—is exactly what the WEF is about. A book co-written by head honcho Klaus Schwab put it thus in 2020 “this transition towards ‘more digital’ of everything in our professional and personal lives will also be supported by regulators. To date governments have often slowed the pace of adoption of new technologies…but…a dramatic acceleration forced by necessity is possible” 27. An aside, it is testament to the damage done by what NFB calls ‘Unofficial Conspiracy Theorists’ (UCTs) that if you mention Schwab/the WEF people think you an anti-vaxx fruitloop. We ain’t—this is capitalism, pure and simple, and we don’t care for it.
Politically, if you think a UK government led by Sunak (a sixth-form student who has wandered into the Head’s office and unaccountably got the job) can regulate AI, think again. The 2023 AI ‘Safety Summit’ held in the UK was akin to inviting vampires to a blood-bank conference. The government doesn’t even grasp what AI is: they only visualise in their ‘Pro-Innovation Approach to Consultation on AI Regulation’ (6/2/24) General & Narrow Purpose AI plus AI agents acting over a longer time-frame 28. They waffle about the rule of law, legal frameworks, an ‘agile pro-innovation’ approach. But haven’t a clue: epitomised by this: future “AI systems may be able to pursue complex goals with limited human supervision, raising questions about how AI agents remain attributable, ask for approval before taking action, and can be interrupted”. What if they are intelligent enough to decide they don’t want supervision, or be turned off, or told what to do, or interrupted? The pat answers (five principles) of Safety, Appropriate Transparency, Fairness, Accountability and Contestability are as much use as an ice-cream franchise in hell. A March 2024 Peking meeting of AI scientists declared “no AI system should be able to copy or improve itself without explicit human approval and assistance” or “take actions to unduly increase its power and influence” 29. Hinting those are AI’s next steps.
Are Labour better? Shadow Technology Secretary Peter Kyle has one big (well quite small) idea—forcing technology companies by statute to share test data, and thereby help the newly established UK AI Safety Institute “reassure the public that independently, we are scrutinising what is happening in some of the real cutting-edge parts of…artificial intelligence…Some of this technology is going to have a profound impact on our workplace, on our society, on our culture. And we need to make sure that that development is done safely”30. Pathetically inadequate, and a dead duck in its own terms—all companies need do is wave the get out of jail card ‘commercial confidentiality’ and, like Starmer’s Labour always does when encountering the powerful, they will meekly back down.
Security-wise we are stumbling (sleep-walking?) into the terrifying situation outlined by Shoshana Zuboff–‘The Age of Surveillance Capitalism’ 31. Of special concern are links between social media outfits/ internet platforms and spooks. On one level, as revealed by Edward Snowden in 2013, spooks are recording and sifting vast amounts of personal data and communications between millions of individuals under an operation code-name Tempora 32. Yet things are far worse: in October 2021 it was revealed “Amazon has reached an agreement with the UK’s three spy agencies to host classified material”. We were also told that “although (Amazon Web Services) was a US company all the agencies data would be held in Britain and Amazon would not have access to information on the cloud platform” 33. Utterly scandalous: as a US-based company of course Amazon will allow the CIA full access, and any idea the company setting up this platform will not be able to hack/infiltrate it really is unbelievable. The pathetic excuse given for this total abrogation of sovereignty is “no UK company can handle the scale of the data” 34—so create one! The line is “Amazon will not have information and sources said sovereignty was not considered a significant issue” 35. It may not be ‘significant’ for our political class who view the UK as a tame US vassal, but two things to consider. If (rightly) Chinese telecoms firm Huawei is considered a security risk because it will share data with the Chinese government, why not apply the same logic to the US? Particularly apposite because at this moment the chatterati are foaming at the mouth at a possible second Presidential term for sex-pest Donald Trump. If, as the chatterati believe, he will introduce the Fourth Reich, complete with Star-Spangled Swastika (and maybe Trump Tower a subverted High Castle in a real-life version of Philip K Dick’s book?), might it not be just a touch worrying that a US-based company controls all UK security data?
IS RESISTANCE FUTILE?
To dispose of a non-starter: the March 2023 ‘Open Letter’ signed (among others) by Elon Musk of all people, calling “on all AI labs to immediately pause for at least six months the training of AI systems more powerful than GPT-4 [Open AI’s Chatgpt]. This pause should be public and verifiable, and include all key actors. If such a pause cannot be enacted quickly, governments should step in and institute a moratorium” 36. It didn’t happen, and won’t—much development is undertaken by ‘stealth’ front companies for a reason, and governments are in thrall to the New Gods of AI anyway. How would it be verified? A pipe-dream, though in Musk’s case just another cynical PR ploy.
Some thinkers and organisations aim to resist. Two from the US are Chellis Glendinning 37, and the podcast/resource library techwontsave .us. On these shores two interesting recent books, a start (who thought Marx’ Capital would ever get traction apart from him and Fred?). The first has a slightly odd title ‘Resisting AI: An Anti-Fascist Approach to Artificial Intelligence’ 38. Resisting isn’t odd, but surely AI isn’t fascist is it? I’ll come back to that. Author/academic Dan McQuillan does good work pointing out problems inherent in what Yevgeny Morozov calls ‘AI solutionism’—allowing inflexible and suspect algorithms to rule the world. McQuillan’s solutions are variously ‘workers councils’ set up by those working in AI from below (p.122) and “broad-based intervention in AI through the mechanism of people’s councils, in parallel with the workplace organising of workers councils” (p.127), all bound together by ‘mutual aid’. Historical precedents include Luddites: good.
Then, McQuillan says “our approach to AI must be an anti-fascist one” (p.133). We are all anti-fascists but not sure how this reductionism helps. The last chapter suggests the “organs of an anti-fascist approach to AI, such as workers and people’s councils, need to position themselves as decolonial through their centring of Blackness in its political sense” (p.137). We are anti-racist too, but I can’t make proper sense of it, and am bemused by the argument “an anti-fascist approach to AI….must be explicitly feminist because of the tendency that AI systems have to enable gender-based violence and promote violently patriarchal systems” (p.137). Yet these activities do not need AI to flourish, they happen anyway. He is really into this AI=Fascism trope, telling Computer Weekly “liberal political structures tend to prefigure fascist ones”, and speaking of “the continuity between the forms of government that we have, and the forms of government that I think are coming which are clearly more fascistic” 39.
I asked our editor to read McQuillan’s book. Larry’s response —“interesting, but large chunks are bowdlerised versions of important thinkers, not named. The ‘mutual aid’ idea was first articulated by anarchist Peter Kropotkin, workers councils have a creditable pedigree such as the Turin Workers Councils (see Antonio Gramsci) and May ’68 in France (The Soviet at Saclay), plus theoretical work by Cornelius Castoriadis, Anton Pannekoek and Karl Korsch. McQuillan’s ‘people’s councils’ recall the original Soviets (in the Russian Revolution) before Bolsheviks crushed them, and organs of Popular Power developing in Chile in the later stages of the Allende government before the 1973 coup. As for anti-fascism, like many on the Left McQuillan uses the term fascist as a catch-all for things he doesn’t like. If serious about this thesis he should have looked at Herbert Marcuse’s ’Reason and Revolution’. Prefigurative politics was key in the 1979 feminist book ‘Beyond the Fragments’ also. But these ideas are not explicitly acknowledged, referenced, or developed, the author has just intellectually copied and pasted them, without historical context, presenting them as a programme which in this ghost form they are not. As for anti-fascist councils centring themselves as decolonial and declared feminism, in modern academia you must shoe-horn such phraseology into everything to get funding: simply declaring yourself anti-capitalist isn’t good for the university bottom line”. Apart from that I think he liked it…
James Muldoon’s ‘Platform Socialism: How To Reclaim Our Future From Big Tech’ 40, unlike McQuillan, places things in full historical context—referencing the Chilean experiment in democratic cybernetics under Allende 41 and mentioning my heroine, Rosa Luxemburg (p.104). He is very aware of the ways capitalism can recuperate oppositional initiatives. Quite rightly, to build the alternative of ‘Platform Socialism’ Muldoon argues for a “threefold strategy of resisting, regulating and recoding digital platforms…this entails bottom-up struggles by workers and users as a method of building our power and eroding the power of tech companies. It also seeks to make use of the existing powers of the state and transnational institutions to fight the monopoly power of Big Tech and curb the worst excesses of their practices. Most importantly, however, we have emphasised a transition strategy…a radical reformism that starts from a realistic assessment of where we are but is not restricted to solutions that appear immediately within reach” (p.136). To be fair, McQuillan probably agrees with much of this, but Muldoon is sharper. The urgent task now is to get workers, the Last Century Left and their descendants to take these issues on board, and develop alternatives from below. A crucial book here is the sadly-neglected (and outrageously priced) ‘Information Technology and Socialist Construction: The End of Capital and the Transition to Socialism’ by Daniel E Saros 42. If/when the Left recaptures Labour (it’s happened before) we need policies in this area.
To paraphrase Morpheus in the Matrix—I have only pointed at some doors to kick in. Let’s get on with it: just maybe German comrades Vulkangruppe already have 43. Replies to this piece welcome.
FOOTNOTES
1) Clay Shirky ‘Here Comes Everybody’ Penguin 2009, ‘Cognitive Surplus’ Allen Lane 2010.
2) Ray Kurzweil ‘The Singularity is Near’ Duckworth 2006 & forthcoming ‘The Singularity Is Nearer’ 2024
3) Nicholas Carr ‘The Shallows’ Atlantic Books 2010
4) Nicholas Carr ‘The Glass Cage’ WW Norton 2014
5) Jaron Lanier ‘You Are Not A Gadget’ 2010
6) Jaron Lanier ‘Ten Arguments For Deleting Your Social Media Accounts Right Now’ Bodley Head 2018
7) James Barrat ‘Our Final Invention’ 2023 Quercus edition
8) The Times 30/8/23 Kat Lay
9) The Times 13/1/24 Tom Witherow and Mario Ledwith
10) as reported in the transcript of an employee ‘all hands’ (ie monologue) verge.com 31/10/23 Alex Heath
11) Momentum Press New York 2010.
12) The Times 9/8/23 Charlie Moloney. See also Sunday Times magazine 8/10/23 Laura Hackett
13) see Notes From the Borderland issue 8 p.33
14) The Times 15/9/23 David Woode
15) The Times 13/4/23 George Grylls, Sunday Times 16/4/23 Tim Shipman/Gabriel Pogrund, The Observer 16/4/23 Julian Borger/Manisha Ganguly
16) ‘Why the 9/11 and 7/7 Cults Bothered Us’ www.borderland.co.uk
17) Sunday Times 8/10/23 Peter Conradi
18) The Times 19/10/23 Fiona Hamilton
19) Aljazeera.com 23/2/24 ‘Google pauses Gemini’s image tool’
20) Tesladeaths.com
21) Barrat op. cit. p.25
22) ‘Why The Future Does Need Us’ forbes.com 26/3/20 Michael A Alvarez
23) ‘AI still has much to learn from how our brains work’ The Times 5/1/24 Rhys Blakely
24) The Times 2/2/24 Rhys Blakely, see also neuralink.com
25) ‘AI to hit 40% of jobs’ BBC on-line 15/1/24.
26) The Times 16/1/24 Patrick Hosking/Richard Fletcher.
27) ‘Covid 19: The Great Reset’ Klaus Schwab/ Thierry Malleret Forum Publishing 2020 p.155
28) HM Government ‘Pro-Innovation Approach to Consultation on AI Regulation’ (6/2/24) Glossary Point 3
29) Financial Times 19/3/24
30) theguardian.com Dan Milmo 4/2/24.
31) Profile Books 2019
32) The Guardian 22/6/13 ‘How GCHQ watches your every move’. See also The Independent 28/2/14 Ian Johnston on a similar ‘Operation Nerve’ regarding Yahoo webcam chats
33) The Times 26/10/21 Nadeem Badshah
34) The Times 27/10/21 Fiona Hamilton
35) The Times 27/10/21 Fiona Hamilton
36) See futureoflife.org 22/3/23
37) ‘Notes Towards A Neo-Luddite Manifesto’ (1990) anarchistlibrary.org
38) Bristol University Press 2022
39) ComputerWeekly.Com Dan McQuillan interviewed by Sebastian Klovig Skelton 1/6/23
40) Pluto Press 2022
41) p.74-76. Also see Eden Medina ‘Cybernetic Revolutionaries—Technology and Politics in Allende’s Chile’ MIT Press 2014
42) Routledge 2014: cheapest anywhere £118!
43) see Le Monde on-line 5/3/24
About The Author
