In which I describe to my 90-year-old father
how the world ends
(probably after he’s gone)
(but maybe before I am)…

I’ve been rereading 1984, and was struck by the parallels between Orwell’s dystopian view of the future, and the development of AI. One “future technology” that stood out was “Prolefeed” (short for “proletariat feed”), the machines that produce the tabloid news, songs, novels, and pornography that keep the steaming masses of the general population entertained and occupied.

In 1984, people aren’t even needed to do the writing, and in any case human input/instructions are captured by dictating into the Speakwrite, and all superseded history is consigned to the Memory Hole (incinerator), so only the officially approved history remains.

Except now we don’t need humans to do the work…AI can (once prompted to do so) create an alternate truth and memory hole the inconvenient historical record without accountability or even intervention from pesky living people to move the workflow along. People are now being removed from the means of production. This may be an exaggeration now, but my lived experience tells me that where profit and pornography are concerned many will keep trying until someone gets it right.

Kind of like a goat testing the fence to the garden – the fence has to be repaired and complete at all times, every time. The goat only has to get it right once.

Then rewriting history, influencing public opinion, or waging propaganda campaigns become even easier. As the cost of erasing “Tienanmen Square” from your national internet becomes cheaper, one can turn that same technology to efforts to project one’s influence to other countries. Tailored down to the individual person level if need be. This would merely an extension of current marketing tools.

China’s efforts to co-opt and pressure ethnic Chinese organizations in Canada, censor sensitive topics, run political interference and influence elections, operate secret police stations, use academic relationships and research partnerships as leverage, leverage economic coercion and inducements, and its role in cyber-operations and ‘information shaping’ in Canada are well known and documented. It could be characterized as “persistent, adaptive, deniable political warfare below the threshold of conflict.”¹²³⁴ Again, AI just let’s them super-charge their efforts, internally and externally, cutting out the high-maintenance humans and replacing it with an intelligence that doesn’t require food or sleep.

But you don’t need to be a major power government to be a player. AI agents are becoming cheap and ubiquitous, easy to hide, and easy to use. Easier, say, than building a nuclear weapon or creating a new virus strain using CRISPR-Cas9. As we’ve learned, uranium refinement is hard to hide.⁵

The people who think about this stuff imagine there are at least two ways that AI can directly bring about an extinction-level event for humans.

The first is by accelerating the division between us. Consider the impact of a ubiquitous, opaque, undetectable, unethical, un-empathetic and competing decision-making and creative forces unleashed for selfish reasons is already having on our local, national, and international politics:

“The rise of unfathomable alien intelligence undermines democracy. If more and more decisions about people’s lives are made in a black box, so voters cannot understand and challenge them, democracies cease to function….When people can no longer make sense of the world….they become easy prey for conspiracy theories….”⁶

Consider how AI can be used in bad faith to accelerate divisions between nations and social groups, much like the mis/disinformation campaigns already being run for fun the last 10+ years by Russia, China, Iran⁷, etc. Except now it’s AI running 24 hours a day, not the human hackers from Slovenia or North Korea that used to get paid by the hour and need to sleep once in a while.

Again, you don’t have to be a national government to have an impact. In fact, it seems to be an advantage to escaping accountability and consequences. Facebook played an instrumental role in the lead-up to the genocide and ethnic cleansing of the Rohingya Muslim minority in Myanmar by serving as a primary vector for the un-moderated spread of hate speech and misinformation, with humans being the drivers as the agents. Replace humans with AI, and you’ve accelerated the generation of hatred. Much like the Chinese ghost factories,⁸ AI doesn’t need to sleep, and it doesn’t need a keyboard, so it is much more “productive” and “effective”.

If that doesn’t seem so bad, remember that Trump was twice elected, supported in part by a Russian disinformation campaign, and last month⁹ he was talking about how the U.S. is going to start nuclear testing again because “everyone else is doing it” (they’re not). 

The second way is for AI to become so integrated in our lives, then deciding it doesn’t really need us. It’s that science fiction trope found in “Terminator”, but not quite as dramatic and with no magical time travelling bad-as Sarah Connor to fix things.

Nuclear option too destructive? Let’s stir together robotics, machine vision, the Internet of Things (IoT) and AI (a la Chinese “ghost factories”), and add existing gene editing technologies to tailor a virus that takes out some, all, or selected bits of humanity. I’m sure an intelligence smarter than us could probably think of some new scenarios, or more likely some comic book or science fiction writer has already given it a couple of good ideas.

“But Bernie”, you say, “we could just unplug the computers. No problem.” Sure, unless that AI is much smarter than us, and therefore much more persuasive. Could it not convince the person or people responsible for unplugging it that actually, “No, you don’t really want to do that, Dave.” Could it not convince us that it wasn’t trying to murder us?

Pandora’s box is opened, nobody can stuff it back in, and there’s no happy fairy at the bottom of the chest floating out farting rainbows and making it all better…we’re going to have to do that for ourselves. 

How do we do that? Isn’t AI good for something? 

---

1. Canadian Security Intelligence Service.
   Public Report 2023. Government of Canada, 2023 ↩︎
2. Public Inquiry into Foreign Interference ↩︎
3. NSICOP — Special Report on Foreign Interference in Canada’s Democratic Processes and Institutions ↩︎
4. Public Safety Canada — “Foreign Interference: Overview of Hostile Activities” (briefing material) ↩︎
5. Iran’s ‘secret’ nuclear program ↩︎
6. Yuval Noah Harari, “Nexus: A Brief History of Information Networks from the Stone Age to AI”, 333-334 ↩︎
7. NATO Strategic Communications Centre of Excellence ↩︎
8. There’s the potential/happening social collapse/upheaval and economic turmoil that comes with millions of former blue and white collar workers being made ‘redundant’. China is already running manufacturing 24 hours a day that doesn’t include lighting because there are no humans on the factory floor. Amazon announced the layoff of 17,000 workers last month (being replaced by fulfillment robots), with probably more to come. Lawyers, doctors, engineers, musicians and artists are also not exempt. This, however, isn’t ‘extinction level’, just ‘major social upheaval’. Like the French Revolution. ↩︎
9. This has been overtaken by events at the time of writing: coup in Venezuela and the impending invasion of Greenland. I doubt nuclear testing quip will even make the footnotes of history at this point. ↩︎