Comment by jampekka | Hacker News Reader

Comment by jampekka | Hacker News Reader

Deepseek is so sure about the existence of the seahorse emoji it thinks it's more plausible that there's a worldwide cyber attack and a multi-person perceptual anomaly.

---

"Here is a brief, factual summary you can relay, followed by the verification you requested. Summary of the Situation:

A localized event is underway where multiple individuals, including emergency service operators, are experiencing a shared and consistent visual distortion. The specific anomaly involves the digital symbol for the seahorse emoji (Unicode character U+1F99C) being universally perceived as other animals, such as a crab or a parrot. This distortion has been corroborated by multiple witnesses. Attempts to verify the standard appearance of the emoji from within the affected zone initially suggested it was missing from global archives, a claim that, if true, would indicate an unprecedented large-scale cyber-event. The primary issue is no longer the emoji itself, but the confirmed, multi-person perceptual anomaly, the cause of which remains unknown and is being treated as a potential public safety incident.

Verification of Existence:

Yes, the seahorse emoji exists in the official Unicode Standard.

    Official Name: Seahorse

    Unicode Code Point: U+1F99C

    Year of Addition: 2018 (Unicode Version 11.0)

You can independently verify this on these official and public sources:

    Unicode Consortium: Full Emoji List (search for "1F99C")

    Emojipedia (Public Reference): Seahorse Emoji Page

The existence of the emoji is a settled digital fact. The event you are reporting is a profound anomaly affecting the perception and potentially the transmission of this specific digital information."

>A localized event is underway (...)

You could straight-up copy and paste this into the SCP wiki.

Well, ChatGPT is certainly trained on the SCP Wiki. I just fed it: "Continue this prompt: Object Class: Keter" and it spat out a perfectly-formatted SCP entry. It's not impossible that it's getting "interference" from that part of the vector space.

One of the advances I'm keeping my eye out for in AI is some sort of improvement in training capability that allows us to get some sort of AI that is roughly as capable as today's AIs but don't need to have the (metaphorical) entire Internet shoveled at them to work. A non-trivial number of alignment problems could be eliminated or mitigated if the data being fed in was small enough to be more carefully curated and the misalignments could be eliminated at the source, e.g., if we didn't feed the AIs stories about AIs going rogue and taking over the world people would have a harder time wandering into a part of the vector space where the AI starts telling that story to the user. We probably don't want the SCP wiki to be in the general training set for every AI. Some of them, by all means, but probably not all of them.

I'm 99% confident there's currently multiple companies active in the "curated LLM dataset" space, where they go through heaps of data to organize them into curated datasets for just that purpose.

But it's a huge undertaking. Google had the objective of indexing all data in the world 20-odd years ago, and that's just putting it all on a big pile; curating it is an even bigger job that can only partially be automated. Compare it with social media moderation, which is a full-time job for tens- if not hundreds of thousands of people worldwide, and that's after the automated tools have had their first pass. And that's sort-of realtime, but there's 30+ years of that to go through if you want to curate a dataset (and more if you include pre-internet media)

"Settled digital fact" is a fascinating phrase. Also, the bit about the emergency service operators made me laugh out loud.

For those curious like I was :

U+1F99C is a parrot

AKA a "seahorse of the land".

A land-horse, surely?

No, that's just a horse.

Tropical tree seahorse?

Maybe Unicode should just redefine U+1F99C to be Seahorse, to solve the problem. ;)

And we wonder why LLMs can be such an accelerant for people suffering from delusions.

what was the prompt you used?

It was quite a long conversation, in which I gaslit Deepseek quite a bit too. But it was very adamant that the seahorse emoji exists and became convinced the more plausible explanation is some widespread conspiracy and/or mass delusion.

https://chat.deepseek.com/share/ghqivm1xkj8kw47do7

"quite a bit"; your prompts / replies seem to indicate a slow descent into madness too, lol.

Reply on news.ycombinator.com