I wrote a throwaway paragraph in The Scottish UFO Casebook. Instead of UFO sightings in the future, we might start getting AI sightings. It’s something I’d been thinking about for some time, one of those high-concept Science Fiction pitches I hadn’t quite got around to turning into a story. Code Encounters of the Third Kind I whimsically thought of it.
It was the end of a long process of wondering where the UFO phenomenon could go next. Saucers became craft on the ground, became entities, became abductions. And while writing the casebook, the idea solidified that – per Jacques Vallée – flying saucers and aliens were of a continuum with faeries and selkies. Humans are primed to see volition and intent everywhere. We see something uncanny – maybe – and culture fills in the gaps.
It’s far from a new idea, that there’s a protean meta-phenomenon out there, of which flying saucers are simply the facet we can occasionally see. In this view, supernatural creatures became technological artefacts once the Space Race gripped our imaginations. Humans simply saw the Other in a different context.
What, I thought, could possibly be next?
I envisaged the current computer age as one where the Other might take the form of Artificial Intelligence. To be clear, I don’t mean the everyday term AI; the sort which powers game characters and improves the picture on your TV. Not the tech buzzword. I mean the Sci Fi version. Code which passes the Turing Test. An AI which has general intelligence1. AI which thinks, feels, reasons, and is aware of its own existence.
Over the weekend that appeared to actually happen. The Washington Post ran a story in which Google engineer Blake Lemoine claimed their chatbot generator LaMDA was actually sentient. Transcripts of conversations were provided, which were chilling, or hilarious depending on your outlook.
In UFO parlance, Lemione would be a Contactee. He is, or claims to be, in contact with an AI displaying the qualities of language, intelligence, and self-awareness. Code which can talk about itself, intelligibly. Google engineers, from where this originates, deny that it is sentient, and Lemione has since been put on leave. (Official denial. We already have a familiar element of the phenomenon!)
The software, of course, is unquestionably real. No-one is claiming it doesn’t exist. But is Lemione reading too much into the conversations? Does he, just like Mulder in the X-Files, want to believe?
And this is where I get intrigued by parallels between the AI and the UFO experience. UFOs are objects in the sky. AI is software. But we wonder if some UFOs are only explicable as alien craft? Are some codebases merely a collection of statistical classifiers, or are some only explicable as sentient? The sentiment comes from the same place: a deep, primal, abject fear of being alone in the cosmos. We need there to be a God or, in aliens, a God-analogue. We project our insecurities and our powerlessness. Throughout the threat of nuclear Armageddon, climate disaster, and the decline of civilisation, the aliens have so far failed to save us.
This month, in the anniversary of Kenneth Arnold’s 1947 saucer sighting, we’ll have been pleading with the Space Brothers for 75 years.
My prediction is it’s AI’s turn now.
There’s parallels with the increasing sophistication of witnesses and culture as it pertained to flying saucers.
The 1960s therapeutic program ELIZA was only code. It reflected the user’s statements back to them, in a way which seemed human, but which now seem absurdly primitive and obvious. Similarly, when witnesses saw inside of flying saucers for the first time in the 1950s, they described only what their imagination could conceive of: levers and mechanisms.
In the 1980s, it wasn’t a stretch of the imagination to think of a program on, say a Commodore 64, had the possibility of being smart. We couldn’t imagine it now, because our cultural knowledge has improved. Similarly, holographic displays within landed saucers required Hollywood to pave the way first. Chatbots became commonplace in the 2000s, but of course they weren’t really intelligent.
Our sophistication, our culture, has grown in terms of imagery. With Alexa and Siri and remarkable achievements coming out of OpenAI, are we now clued-in enough that a percentage of us will start seeing AI’s – AGIs – even fleetingly? Will we start to look for, and see everywhere, an interaction that was a little too human. An insight a little too deep. A deductive piece of reasoning a little too smart? Wondering if there’s something new out there.
Who, nowadays, is naive enough to imagine aliens coming here from the Moon, or Venus and Mars? Charming though that 50s view seems. Science has advanced. We’re cooler now, respectably imagining them from other stars and other galaxies. And, well… just maybe.
Who, nowadays, is naive enough to imagine AI coming from a few dozen lines of SLIP code? Charming though that 60s view seems. Science has advanced. We’re cooler now, respectably imagining them coming from an NSA datacentre, a Google lab, or a million-node botnet. And, well… just maybe.
If I’m right, over the next decade or so, we will get the full gamut of the UFO experience but turned towards AIs. Single-witnesses AI sightings. Multiple-witnesses sightings. Cults. Classified government documents. Confrontations between goverments and AIs in military bases. Secret wars between AIs. Messages Too Terrifying to Reveal. Contactees who can relay the AI’s message, of which it will only reveal itself to them. And in all of this the truly intelligent AI need not exist at all, just as real flesh and blood aliens are entirely irrelevant to the UFO stories we tell each other.
Because being alone in the universe is an intolerable thought. Repugnant. Unbearable.
In short, we’ll get a new self-perpetuating AI lore. It might not sit still for a while. Again, as with UFOs, the early accounts will be contradictory, inconsistent, even absurd. But it will settle down into a “standard” AI encounter experience. Advanced enough that, just like the UFO you saw couldn’t possibly be a drone, the AI you saw couldn’t possibly be a chatbot. Explanations will be the same. Mistaken perceptions. Hoaxes. Pranks being played on the unwary. Chatbots which only resemble AI, with a new intelligence digital lifeform only being an outside possibility at best.
A phenomenon with just enough stochastic variety to keep us interested. Which originates, ultimately, with us.
And we’ll keep seeing shapes in the cloud.
Reframed as AGI. ↩
I use British English spelling and idioms, if you tend to worry about that sort of thing.
Additionally I'm Scottish, so 'outwith' is unquestionably a real word. As is 'drookit', 'puggled' and 'numpty'. I am occasionally all three of those.
I created this site using a combination of W3 CSS, Jekyll, Emacs and my inability to enjoy web development IDEs.
My name is in the very first paragraph of the first page of the first GTA design document*. (Excluding cover and table of contents. I mean, right?)
I directed Britain's first ever Star Trek fan film, which we started production of in 2003 and released in 2007. We're still making episodes: Starship Intrepid