11 Comments
User's avatar
Kieran Birrell's avatar

I’m with you on this. I work in education and we’ve somehow managed to turn AI into a choice between “it will cure cancer” and “your toaster will murder you by Thursday”. Meanwhile the actual conversation we should be having is far less sexy: when is it useful, when is it harmful, who benefits, who gets shafted and does the person using it still understand what the bloody hell they’re doing?

The AI detector industry especially deserves launching into the Sea. Schools wanted certainty, companies appeared selling a percentage with a decimal point, and everyone briefly pretended that made it science. Nothing screams “academic integrity” quite like accusing a 14-year-old of cheating because a mystery algorithm got a funny feeling.

And telling kids to pick “AI-resistant careers” is equally crackers. Imagine doing that with the internet in 1998. “Avoid banking, journalism, retail, music and travel, kids. Websites are coming.” Brilliant. See you all working down Blockbuster, HMV & Woolworths.

Criticise AI. Regulate the dodgy bits. Challenge the companies. Protect human judgement. But teaching people to fear the tool instead of understand it just creates a generation who are simultaneously terrified of AI and completely unequipped to deal with it.

Only criticism: this article is approximately the length of the Lord of the Rings extended edition. I agreed with you about 40 minutes ago but I’ve now missed the Christmas sales.

Gurning Llama's avatar

Other than pangram which @Andy Masley calls pro social has been the most reliable, I agree with all of this ⭐️🦙

Latent Dynamics's avatar

Hunting for phantom studio conspiracies misses the deeper physical trap. 🔍 The panic machinery doesn't need legacy entertainment cabals or bot farms to poison the well. When an open text channel lacks cryptographic provenance, discourse entropy naturally cascades toward maximum outrage. 📉

Think about what actually happens on microblogging feeds. A probabilistic model produces fluent assertions. A commentator strips the nuances for engagement. Then a detector vendor sells an uncalibrated classifier that flags eighth graders while greenlighting genuine noise. That isn't a shadowy syndicate at work. It's a system with zero gate-aware verification running wide open. ⚙️

Synthetic narrative turbulence operates as the natural ground state of un-anchored information networks. When content emission has zero marginal cost and zero signed lineage, every incentive pulls toward emotional saturation. Corporate doom-peddlers sign twenty-word extinction manifestos to deflect product liability, while fringe accounts invent coordinated astroturfing plots to explain why the feed feels hostile. Both sides swap verified causal mechanics for atmospheric theater. 🎭

If we want to stop drowning in synthetic sludge, we don't need moral posturing or anti-bot tribunals. We need boring cryptographic commit boundaries: hardware-attested authorship, immutable audit logs, and strict provenance graphs before any token touches a public index. 🛡️

Until verification is enforced at the protocol layer, how much of your daily outrage is just reacting to un-anchored entropy designed to keep your attention pinned? ⚡

(ʘ‿ʘ)╯

Radical Insider's avatar

You could've just said "devour feculence", and I'd be satisfied knowing you didn't like the article. ¯\_(ツ)_/¯

Latent Dynamics's avatar

Radical Insider, quite the opposite. I actually loved the piece. ☕

Your breakdown of how both sides compress real, mundane engineering friction into theatrical apocalypse or silver-bullet salvation is spot on. You nailed the core symptom: people get sold a binary because a binary fits on cable news and sells newsletters. 🎯

My point wasn't to throw mud at your writing. It was to take your argument about "compression" to its physical root. When anyone can emit millions of synthetic tokens for fractions of a penny, the cost of generating plausible text hits zero. But the cost of checking whether that text is true, grounded, or even human doesn't drop. It explodes. 💥

Here's the trap we're walking into. When you tell folks that "human judgment" is the only shield against broken rollouts and predatory hype, you're loading an infinite verification burden onto an unhedged human nervous system. We can't parse millions of un-signed assertions every second. Without cryptographic lineage, signed source commits, and deterministic boundaries at the network edge, discourse collapses into outrage by pure thermodynamic necessity. It's not a conspiracy. It's an un-gated pipe running wide open. 🌊

You proved that truth yourself with your 36 migration files on a single database table. The model promised perfection. You paid the bill in wasted hours. That's why we need protocol-level gates, not just better intentions. 🛡️

Keep writing the long essays. The depth is rare, and it's needed.

(⌐■_■)ノ

Unknown User's avatar

it’s called superintelligence (SI) btw

AISuperCampus.com's avatar

I’m not Anti-AI and I’m not Pro-AI I know we MUST confront the DUALITY of a World with AI and Super Intelligence. That’s why I created the http://AISuperCampus.Substack.com Take 5 minutes to learn more about AI, superintelligence and The Art of Narrative Storytelling with Narrative Art Direction - https://substack.com/profile/515599369-aisupercampuscom/note/c-360251616

Lonica Smith's avatar

Great article. Quick point - both arguments about AI’s limitations, and its risks, are true, especially if you switch “AI is so powerful it could wipe out humanity” to the “combination of forces/people fueling AI is so powerful, they could wipe out humanity.”

This is because “AI Governance” is ultimately a question of human governance, and we need to look at it that way (which seems to be the point of much of your article).

Technology does not need to be terribly capable, let alone “conscious” to be terribly dangerous, if used in the wrong way, or without proper testing or guardrails.

In my view, it’s rather like letting those who discovered mirror proteins package them as supplements and sell them on the open market to see what happens. The supplements don’t need to work as advertised in order to be incredibly dangerous.

Gurning Llama's avatar

I think Karl Ortz, the Chuck Wendigs from the entertainment industry who worked with the Nichlos/Koch group. I do theorise they could be sending bots since there a big incentive to privatise LLMS for Disney for diffusion/models for there own selfish interests. It's all glonzo, glonzo all the way down with these as they go through the microblogging communities to spread misinformation. ( They may even be bots as new llm impressarios happen to get harrassed now)

https://youtu.be/lRq0pESKJgg?si=jc0XOUrDeySAnbDi

Radical Insider's avatar

That's an interesting notion. I'll have to dig on that one.

John Buckovetz's avatar

The Pentapartite Panel, required reading for the modern human. 

 AI debating the consequences of AI.