Hello,
Last week I was at the glorious Nickel Cinema in Clerkenwell, London, to screen a short film project that I’ve been working on - on and off - over the past 4 years. It was an event organised by Jonathan Baggaley, editor of Fully Human, the creative research and development arm of the PSHE Association. For those who aren’t acquainted with PSHE, it’s the Personal, Social, Health and Economic education programme in UK schools - what, in my day, was called PSE, which was optional and tended to consist of your form tutor looking mortally embarrassed as they put a condom on a cucumber. Things have developed a lot since then, in no small part thanks to the Association, which successfully campaigned to make it mandatory to guide young people through the multifarious challenges they face in an increasingly complex and confusing socioeconomic landscape. It’s serious business.
The invited audience was made up of leading thinkers in the fields of education, critical AI studies and child online safety - people like Future Narratives Lab, who do amazing work in the field of systems change, and the founders of We & AI and Better Images of AI, who are building a vital public knowledge-base for AI literacy - and they were brought together by Jonathan to have a frank discussion about the issue of anthropomorphism in AI and its role in the Attachment Economy. The film was there to kick off the event and frame the discussion, and I will admit to being slightly nervous about it, not least because it’s deliberately lighthearted despite being deadly serious about its intent. Would these serious people get it? Would they care?
Jonathan and I first began chatting about this subject in around 2021, after I completed my first film as part of my MA at the RCA (a project that I still haven’t found a comfortable place for online, partly because it’s related to the surveillance capitalist capture of the internet itself). Jono had seen first-hand the dangers of the internet to child safety through his position as the Head of Education at the Child Exploitation and Online Protection centre; I was convinced that the tech industry’s obsession with fusing human consciousness with the digital network was leading us somewhere potentially disastrous, and my research for So Long (We Dreamt of This) interlocked with Jono’s understanding of the subject such that we began meeting regularly to chat about latest developments.
This led to my involvement with Jono setting up Fully Human, which initially set out to run workshops using physical media to investigate young people’s growing relationship with extractive and coercive digital platforms and tools, as a means to help them maintain agency within an environment increasingly governed by them, but soon this shifted focus to voice-assisted surveillance technology and the ways in which it leveraged anthropomorphism to influence users’ behaviour, emotions, and navigation of their environment. We went into a handful of schools to pilot the workshops, teaching classes about how AI systems work and how the gulf between what they actually are and what they output is a latent space from which all manner of delusions can be fostered unless you are fully cognisant of its mechanics.
This was in 2022, and during the week of our second workshop in schools, the first iteration of ChatGPT was released. It was no longer something we could say ‘but imagine if…’ about; it was here and we had to deal with it. Given that, as we’ve now seen, even AI engineers and prominent scientists fell into the AI delusion trap (the belief they’re interacting with a consciousness), it seemed even the most educated on the subject were not safe from the designed deception that determines the efficacy of Generative AI.
Despite this - or maybe because of it - we decided to put our thoughts into the form of a video essay, at the very least as a kind of ‘thinking assistant’ to remind us of our position within a truly extraordinary degree of noise, shape-shifting and goal-moving. It was clear to us that no matter what new developments occurred, this issue would always be relevant. So while there are points in there that I’m sure most of us already know, and use of public GenAI tools like NotebookLM and Suno that are already old hat, it still acts as a marker, a reminder, and a kicking-off point for further discussions.
Synthesising this in the form of an audiovisual essay proved very challenging. This is something we care very much about, and believe it to be fundamentally important, but, being in the realm of phenomenology and semiotics, it will probably always be deemed less urgent than issues of state security or environmental harm. Psychological manipulation through attachment techniques will always be deemed so complex and nuanced by some as to be too difficult a subject to tackle. Despite it being clear to most that smartphones and the apps are addictive by design, many still argue that it cannot be technically bundled in with other types of established addictions. But for me (us?), the human need for connection and the platforms’ multifarious techniques for simulating that connection - culminating in chatbots, voice simulation and deepfake technologies - is the primary hook, and as such it is a mechanism that should remain at the centre of debate. This also came from my view that AI is an animation-adjacent technology, and new media theory is a valuable lens for looking at it. We had to hold strong in our commitment to not fall into the twin sins of doomerism and boosterism, to undercut the pomposity of the AI CEOs, while simultaneously making the case that this is still worth thinking about.
In terms of tone, it was essential that it was playful, humorous, mischievous and deceptively complex in order to reflect a certain kind of human creativity and associativity that AI cannot replicate. We needed it to feel very different from the AI-generated videos in order to make our point, and yet we also needed to incorporate AI generated material to demonstrate how difficult a landscape it was becoming to navigate. We wanted it to reward repeat viewing, but not overload on first watch.
In part 2 of this post, I will cover some of the processes and techniques that tackled these challenges, including experiments that didn’t make the cut.
You can watch the video here in its full 7m30s - I hope you enjoy it, and I’d be very interested to hear your human thoughts: https://www.youtube.com/watch?v=50zqcId4KX8
Thanks to Jonathan, Alexandra, the Fully Human team and The Nickel for making it such a great screening - it was a very unusual and unique event of the kind I’ve not experienced before. The post-screening discussion was rich and still going after an hour, the point at which I allowed myself to think, okay, that went really well. Special thanks also to Loïc Kreyden for the help on sound design for the film, and Lucy Ellis for the beautiful creative coding animations that really brought the dataviz sections to life.
Back soon with some behind the scenes,
Lewis
This is a public episode. If you'd like to discuss this with other subscribers or get access to bonus episodes, visit lewisheriz.substack.com/subscribe
