BlueMonday1984 @ BlueMonday1984 @awful.systems

he/they

---

Read more

Posts

100

Comments

1,177

Joined

2 yr. ago

Tried following that link and it took me to a blank page, here's one that isn't broken.

Someone tried Adobe's new Generative Fill "feature" (just the latest development in Adobe's infatuation with AI) with the prompt "take this elf lady out of the scene", and the results were...interesting:

There's also an option to rate whatever the fill gets you, which I can absolutely see being used to sabotage the "feature".

New Public Good newsletter, talking about YouTube editing users' uploads without their permission/knowledge.

I’m now realizing most programmers haven’t done a manual labor task that’s important. Or lab science outside of maybe high school biology. And the complete lack of ability to put oneself in the shoes of another makes my rebuttals fall flat. To them everything is a nail and anything could be a hammer if it gets them paid to say so. Moving fast and breaking things works everywhere always.

On a semi-related sidenote, part of me feels that the AI bubble has turned programming into a bit of a cultural punchline.

On one front, the stench of Eau de Tech Asshole that AI creates has definitely rubbed off on the field, and all the programmers who worked at OpenAI et al. have likely painted it as complicit in the bubble's harms.

On another front, the tech industry's relentless hype around AI, combined with its myriad failures (both comical and nightmarish) have cast significant doubt on the judgment of tech as a whole (which has rubbed off on programming as well) - for issues of artistic judgment specifically, the slop-nami's given people an easy way to dismiss their statements out of hand.

It would be extremely funny if your prompt injection near the end actually caught someone using the Comet Assistant.

(It probably won't, given the Venn diagram of "Comet assistant users" and "Pivot to AI readers" is two circles places a mile apart)

No, I do this for the love of the game

New Ed Zitron: "How to Argue With An AI Booster", an hour-long read dedicated to exactly what it says on the tin.

The Trump administration could've gotten some rando on neocities or nekoweb to do their website and unironically gotten a better result than this bland garbage.

The favicon is just 16x16 pixels of the word “by” in cursive that’s so blurry you can’t actually tell that’s what it is.

They might as well have gone with the Schutzstaffel lightning bolts - they're pretty recognisable even if the resolution is Jack x Shit, and they fit Trump's general ideology pretty well.

Thinking about it a bit, I suspect you're not alone - whilst the '00s were pretty great for me (I was born in 2000, remember), the '10s were a complicated mess (for a long list of reasons), and the '20s have been one wash after another - and thanks to the 'Net, I'm aware how much hot garbage the '00s and earlier decades had.

I do also have individual bits of media which I've got fond memories of, but that's about it. Thinking about it, my general soft spot for Y2K stuff is probably a lot less rooted in nostalgia than I thought.

No idea. Best guess is that its some attempt to sell fascism under a thin veneer of '80s nostalgia, mainly because this one full minute of Oops! All White People, and fascists are the biggest boosters of AI bar fucking none.

When it comes to falling for nostalgia, its generally Y2K-tinged stuff that gets me - I'm much closer to that era (again, I was born in 2000), and I've got a soft spot for the general visual style of that era (that its facing against Corporate Memphis/AI slop definitely helps).

This is a shot in the dark on my part, but I get the suspicion the quantum hype is gonna face direct resistance, in a similar manner to LLMs/AI.

Quantum's supposed encryption-breaking abilities are currently a hypothetical, but hype about such abilities could prompt fears that governments/corporations would abuse quantum to supercharge currently existing mass surveillance, enabling governments to invade people's privacy without needing a backdoor or corporate cooperation.

High energy consumption will likely prompt resistance as well - the current crop of quantum computers consume a lot of power to keep their chips within spitting distance of absolute zero, and after seeing AI corps do everything in their power to consume as much energy as possible, I can see the public expecting similar behaviour in the upcoming quantum bubble, and reacting accordingly.

On a wider front, I expect machine learning as a whole will see its funding go down the shitter once the AI bubble bursts - useful ML applications are losing funding in favour of magical chatbots, and the stench of AI is definitely wafting all over it as well.

For an off-the-cuff prediction, I expect the number of AI/ML researchers to steeply drop in the coming years, from funding for AI/ML drying up post bubble, and from researchers viewing the field as too tainted to touch.

The flag was the most obvious one I could think of, given how many eyes were already on HWNDU and how swiftly they found it. In retrospect, I should've chosen 1d4chan/1d6chan as my example, given how large and robust it is as a wiki.

The SCP Foundation arguably qualifies as well - it began on /x/ as a random post, before morphing into the ongoing collaborative writing project we all know and love.

Even 4chan can trade/coordinate/and have functional outcomes, sure often for evil.

To give a rather notorious example, there's the He Will Not Divide Us flag in 2017, which the 'channers tracked down after only 38 hours, despite Shia LeBouf's attempts to keep the location hidden.

The death penalty of not just you but your whole family if you copy that floppy.

The future media conglomerates want. (okay maybe not the "death penalty" part - dead people don't make money)

In other news, I've stumbled across some AI slop trying to sell a faux-nostalgic image of the 1980s:

Unsurprisingly, its getting walloped in the quotes - there's people noting how it misrepresents the '80s, people noting much the '80s sucked and how its worst aspects are getting repeated today, people noting the video's whiter than titanium dioxide, people suggesting there's suicidal undertones to it, and a few comparisons to San Junipero from Black Mirror here and there.

Personally, this whole thing has negative nostalgic value to me - I was born in 2000, well after the decade ended (temporally and culturally), and the faux-nostalgic uncanny-valley vibe this slop has reminds me more of analog horror than anything else.

Here's my idea to increase the birth rate:

Make the world less of an all-consuming dystopian hellscape, so people can actually start and raise a family without ruining themselves, and can feel confident their children won't have horrible lives.

Ran across a viral post on Bluesky:

Unsurprisingly, the replies and quotes are universally outraged at the news.

It feels like the rise of LLMs has set back cybersecurity by a good decade or so, and by my guess it probably has.

Agents are throwing away decades of hard-learned lessons in input sanitization (providing cybercriminals a Greatest Hits compilation of vulnerabilities), "vibe coding" is introducing vulnerabilities aplenty to codebases and hiding them under mountains of technical debt/unmaintainable code, LLM usage is damaging coding ability in coders both junior and senior, the entire tech field is haemorrhaging talent from burnout and layoffs, and that's just the things that are immediately coming to mind.

As I see it, cybersec may find itself practically back to square one once the dust settles.

New article on AI scraping just hit The Register, with some choice quotes from Anubis dev Xe Iaso. Xe herself has given some additional thoughts.

On the one hand, I can see your point - such advertisements could provide the hucksters some positive spin to assist their bubble with.

On the other hand, Silicon Valley's still got the heavy stench of Eau de Asshole off the AI bubble, with some Eau de Fash off of the Trump administration. If a new tech can be used for evil shit, the public's gonna (rightfully) assume it will be used for evil shit - and I doubt the hucksters can convince the public to think otherwise.

(also ew, mobile Wikipedia)