iHuman

There are at least two dissonant strands within iHuman and they contribute to its weaknesses. There is a credible and concerning element about how a suite of technologies that it broadly lumps under the heading of 'AI' may be, can be, and are being used to oppress and exploit. There is also a millenarian fantasy where the letter working hardest with 'I' is not 'A' but 'F'. "IF" has rarely been called to shoulder as much since Kipling.

Hidden within iHuman is an innovation that has powers and abilities akin to a person, but is artificial. However there's never a discussion of corporations, nor of governments, at least not as innovations in and of themselves. There are concerns about specific named entities - Google, Palantir, the People's Republic of China, the Greater German Reich - but in and amongst a significant speculative quantity there's never an open discussion of the harms imposed by notions like 'shareholder value'. Indeed, so entrapped by a particular mindset does iHuman seem that it credits two of its talking heads as Angel Investors.

These talking heads are not circular wheels studded with eyes, but people looking to make money. There's a point where one of the multitudinous opiners talks about firms going after military and law enforcement contracts. There are no interlocutors here, heads address someone off camera, but had there been there's a question unasked - is, perhaps, the reason that firms based in the US chase these things that, as bank robber Willie Sutton apocryphally said about his targets, "that's where the money is?" In its attempts to prioritise concerns about automation and to elucidate the risks in potentia, iHuman seems again and again to miss actual things happening on the ground.

This despite a good section on the systems of imposition upon the Uighur minority, a comment from an EU commissioner about her fears drawn from history. Yet a historian might have been useful - ultimately algorithmic bias and mechanical oppression are an addition to the vernacular of authoritarianism - that a curfew or a voting test is administered by a machine and not a notionally biased individual is an attempt to give things the wash of objectivity. Which is potentially ironic, as my subjective judgement will be encapsulated into a number of stars, which can be weighed and measured against how I have judged other films. Yet just like that titular 'i', you cannot remove the imaginary when it is there by itself. You need another to even get to the negatives.

Which iHuman has, sadly, aplenty. Ignoring the sudden descent into stroboscopic dubstep, or the unfortunate consequences of home viewing which mean that artistic imposition of pixelation on footage had me suspecting some issue betwixt WiFi, Vimeo, HDMI and my eye, or the wiggliness of its theses, there are unchallenged statements in iHuman that are not credible

There are multiple, competing, definitions of AI within the film. There's algorithms, predictive policing, imposition of systems, but there's also the fantastical 'AGI' where the G is 'General' and not 'Gullible'. Suggestions of Earth covered over by data centres and solar panels have shades of The Matrix pulled over ones eyes. A suggestion that the first AGI will be better and smarter at any task than a human, and use perhaps as much electricity as 30 million houses. With an AGI available, why would anyone hire a person to do a job? Arithmetically, I can think of perhaps 29,999,999 reasons...

It was some 93 minutes in before I registered the word 'governance', and ultimately that's where a lot of the issues iHuman raises exist. Systems do not exist in a vacuum. A test that you can pass if your grandfather voted is no more or less fair than the system that predicated your vote upon it. There's discussion of the radicalisation of many on the Internet, but no parallels drawn with the bierkellers where groups talked in isolation and unemployment. There's, without warning, footage of Charlottesville. Part and parcel of this collage - I counted 104 credits for "YouTube sources" or equivalent.

With the overwhelming and unspoken issue being the influence of money, the presence of some 20 funding bodies in those credits isn't perhaps ironic. What is, perhaps, are comparisons to the development of the nuclear bomb. For all their fears it was a much shorter path between the work of some physicists and the Manhattan project. While their Fat Man and Little Boy left an unequivocal mark upon the world and an irrevocable influence upon the iron around us, 67 cities other than Hiroshima and Nagasaki were extensively damaged without an atom being split. But for a different kind of cloud it would have been Kokura on that list, but even without Oppenheimer at al that list would have existed.

Which is I think the issue with iHuman - that half of it overlooks the fact that lists exist, and the other half is asked to talk about the influence of 'AI' upon that list. Ignoring that AI as a definition evolves as quickly as something half-baked and then left to decay, these are disparate directions. There's talk of Cambridge Analytica and audience tailoring, and only some of how audiences are themselves adjusted to better fit the ads. There's a suggestion that this kind of targeting is something politicians cannot do, which must be news to any statesperson who's been accused of kinship to Janus. The wooliness around what it calls AI covers everything algorithms to facial recognition to exponential singulatarian super-genii (and I use that plural advisedly, and if you've the bottle you'll see the implication). The fact that it's got footage of a tiki torch talking about great replacement while it's also got someone it tells us was hired by Elon Musk worrying that AIs will supplant us may be presented without comment. I can't let that lie.

There are moments where there's the opportunity for old ideas - one statement around the notion of a social credit score is that it's a system with "No way to challenge the designation or to find out they've been put in this category". That's the automation of privilege, the encoding of racism. That's a new routine on the old spin - objectivity-washing. "points-based immigration system", "personal improvement plan", two stars out of five, that's all opinions (some before the fact) coded (term used advisedly) such that they give cover to biases, implicit, explicit, or otherwise. Someone worries that these technologies could fuel an eternal dictatorship, and that is perhaps true. We've had industrialised and systematised oppression before, however - the issue is not the technologies but the humans behind it. Lessons from history may go unlearned, but no historians here. More's the pity.

There's a, and I'm at pains to be careful here, discussion of machine-learning that can recognise 'homosexual faces'. There's a later reference to phrenology explicitly, from the same head talking. Just because their means were unsound, we're told, and their intent was discriminatory, "doesn't automatically invalidate the claim". Let's not go into how accurate the ability of a photographic system to identify self-identified same-sex attractant individuals on the basis of similarity to a notional set of positional characteristics where those individuals appear to be of the same race and class and socioeconomic circumstances (you can sort of spot a student, can't you?), where there might be some grounds for suggestion that there are reasons people who are sort of the same might look sort of the same - the suggestion there's also perhaps an 'atheist face' takes us into a realm of phenotypic expression of biological essentialism that probably ought to invalidate the claim. I often use a particular discussion of the complexity of class signifiers which is that if one breaks down in the Highlands and, outwith range of mobile telephony, chaps the door of a wee gatehouse and it's answered by a man with snaggle-teeth and a torn jumper and salt and pepper hair and a rabbit in one hand one doesn't know (and even then for certain) if it's ghillie or Laird until he speaks.

There's talk of data as if it's the new currency, and it might be, but without interpretation it's just noise, not signal. Which is true across iHuman. The music is good, Olav Øyehaug does good work, behind the camera Henrik Ipsen has lensed at least as many documentaries as Olav has scored. There are shots though that are just Hong Kong, and I'm not sure what the second system in the One China is meant to convey other than a sense of futurity, but not perhaps real futurity, the echoes of the past. Maybe a historian could shed light. Sadly, director Tonje Hessen Schei has not. Her film wanders around the topic with an irritating mix of credulity and conspiracy - more annoying because there is undoubted malfeasance. Yet it never seems to get near the root of all the evils it's exploring - money, and behind that, power. There's talk of sums of money and google (US$9m, US$250m, US$10B) but never, for context, Google/Alphabet's market capitalisation (US$1.17T). That's T for Trillion. Which makes those three sums (in total!) a mere 0.87%. Don't get me wrong, it's chunkier in comparison to an annual turnover of US$160B and change, but what compromised "don't be evil" is multifaceted.

There's talk of (the notoriously litigious, albeit by proxy) Peter Thiel, and discussion of how Facebook and Google are ad-brokers. Not of how their information is what newspapers used to have, though some discussion of how they too can be used to influence. There's a discussion of suppression, of echo chambers, but there's no real depth here - even if some computer generated sequences seem to be itching for 3-D.

Specially screened for the UN, and released on Human Rights Day, iHuman is worthy, and good, but not worthwhile, nor very good. It undercuts its concerns about the digital architecture of authoritarianism with sequences where a breeze on the wet finger of prognostication inevitably implies the whispering breath of the heralded God-Machine whose coming shall see all other life supplanted. There's a bit where I saw a copy of Hofstader's 'Godel, Escher, Bach: An Eternal Golden Braid', a 40-odd year old text whose 777 pages explore parallels and symmetries and intelligence and more. Its spine appeared unbroken, and that suggests it was unread. There are ideas that are unexplored, others that are elided, I think that there's a point where the question is actually begged, and while there's discussion of ethics and automation the bigger issue is ethics itself.

There's a quote attributed to Bill Moyers - "I'll believe corporations are people when Texas executes one." There's a counterargument that bankruptcy might count, but that way lie zombie football teams and that's not the right kind of ranger. The heart of the quote is violence, however. Corporations undoubtedly do it, and states (small s) spend money upon it. There's mention of evolution and so to return to a point iHuman glosses over.

Let me crib from Khan Academy - the seven criteria for life, but as applied to the modern corporation.

Organisation - specialised coordinated parts. HR department? Metabolism - probably, energy consumed or produced, complexity created or removed, even if only financial Homeostasis - preservation of a status quo, let's say shareholder value Growth - check, if not unchecked Reproduction - persistence of the entity aside, new divisions, mergers? Response - "irritability", so yes. Evolution - often used in iHuman, though perhaps as colony entities heritability is a different trait?

Nonetheless - corporations are more alive than anything purported to be AI within this. Undoubtedly capable of malice, but born from their creators.

Which leads me, I suppose, to that point about the persistence in arithmetic of the root of minus one. iHuman - no matter how much humanity is discussed, the imaginary element is still there. Similarly, no matter how fantastic its suppositions, the human element is still there. What does that all add up to?

One to miss.