“Write to the market.” You’ll never catch the market. By the time you’ve written it, the market is gone. Write what you’re on fire to write, and sometimes the market shows up.

Saying it again, anthropomorphized "chatbot" user interfaces to LLMs should be illegal. There's zero reason these interfaces have to be like this—other than companies making people addicted to them.

Whenever someone suggests lore matters at all, when the point of storytelling is dramatics, I get annoyed

y'know why The Raid rocks so much? because we only know a few important things about Rama (he's a rookie with a wife, kid, and estranged brother) and then puts him into the most insane, unimaginable situation possible what matters is the dramatics, NOT A SINGLE CHARACTER'S BACKSTORY

Firefox browser falls to AI. What do we do now? where ‘absolutely not’ means ‘maybe later’ www.youtube.com/watch?v=c40i... - video pivottoai.libsyn.com/20251223-fir... - podcast time: 9 min 39 sec pivot-to-ai.com/2025/12/23/f... - blog post

Some of you have forgotten that only three years ago you were perfectly capable of writing a text, writing an email, telling a bedtime story to a child, and it should worry you that powerful companies have convinced us we can’t do things we’ve been doing since the dawn of time.

Tonight Kertasníkir (Candle Stealer), the last of Iceland's 13 Yule Lads, comes down from the mountains. He steals candles (that used to be made from animal fat) and eats them. Children leave their shoes out in the window, he gives the good ones a small gift, the bad ones get a potato.

How is this legal? And if it is legal, what exactly do we take a torch to it legislatively? www.washingtonpost.com/lifestyle/20...

You know, we're always told we view the past with rose-tinted glasses & the olden days were never really that good, but I remember when journalists used to piss & moan about having to do human interest stories instead of hard news because there wasn't enough news to fill half an hour. I miss that.

This is the gist of what I’ve heard from people who do use AI / LLMs at work (outside actual data handling). Its best workplace use is to automate stuff that nobody should be asked to do in the first place. Automated time-saving generation of the things no human ever looks at.

love that software developers are advertising AI functionality in every single app now and it's up to you to try to figure out if they mean "this has spellcheck" or "everything you create in this app will sync in real-time to sam altman's personal laptop" or something in between

by far the funniest use case for AI is “I use it for powerpoints because I hate making them and nobody looks at them” because it kinda sounds like the real workflow savings would be to stop making powerpoints

Very good paragraph here from @cstross.bsky.social on LLMs… www.antipope.org/charlie/blog...

This is a wonderful summation of the beauty and value of human made art ❤️

One of the things I’m trying to figure out with my impromptu photography retrospective over the past few days is a better understanding of how I can deal with colour. I’ve been trying to figure out when colour is necessary, when it adds, and when it detracts, how best to use it. 1/6

We begin, appropriately enough, at the end. Where everything is revealed and some things are put to rest. #12daysofYule #thedevilssound #witheredhill

PREACH IT TECH JESUS

bsky.app/profile/pook... The einstellung effect has been known for generations: you have a very hard time noticing you're not actually solving the problem you actually have, but a different problem. Generative AI is the einstellung effect reified.

Learning how to pay attention to small details, how to deliberately strain yourself to notice tiny details that are off, is a skill programmers learn the painful way. And, using a chatbot deskills you in this very extremely important skill of being able to pay attention to small details.

hours of tracing an error only to bolt awake in the middle of the night - your subconscious having clocked away to find a solution. denying yourself those moments of joy and discovery leads only to copy-paste limbo with zero innovation

as a black tech anarchist who hates closed-sourced applications and the concept of BIG TECH, the last 5 years of my existence have been so validating for me

Somewhere out there right now is a programmer vibe-coding an FDA-regulated medical device, and somewhere else is the unemployed person whose job was to inspect that code and hold it to standards and who was fired by a 19 year old.

Also, this information isn't in any training module... cuz I haven't published it yet! And there's no other way to learn this except in the archives where I read the panicked letters between Lougee and the IT department! There is *so much* history AI does not know! 🗃️

It's good that more attention is being paid to the plague of AI-drive scams that is driving writers batty (but please, don't mention Writer Beware, which has been reporting on them since August) www.thebookseller.com/news/authors...

obligatory plug of this piece @kara-williams.bsky.social and I wrote a few months ago explaining the stakes of policymaking with vague/improper ideas of what AI means consumerfed.org/specific-ter...

I believe an entry-level software developer's learning journey should include diverse outside influences because it would be very bad luck for about 90% of them if it didn't. "We want them to learn how we do things here." And then, hopefully, they get to learn better ways than that.

This is why @emilymbender.bsky.social @alexhanna.bsky.social say to name the specific thing rather than calling it "AI". Right now, ChatGPT is bucketed under the same term as a model train to translate one specific language into another and which is carefully trained through curated data.

Wildly different things, tasks, techniques, subspecialties being lumped into "AI" and then being conflated with each other, doesn't help. Different types of models vs the techniques to train them vs the tasks they are supposed to accomplish, all under "AI".

Look up a graph of OpenAI's incestuous funding. If you and 6 friends passed the same $20 back and forth, none of you would get rich. But when it's 7 billionaire freaks passing the same wad of cash back and forth, somehow they do get richer. Even while losing billions a month doing this.

"ok so the plan is I buy my couch for $100 and then sell it back to myself for $100. If I do this several times a day I will become rich" But when a billionaire does the exact same thing, somehow it actually works and makes him even richer. Damnedest fucking thing.

It's funny how if *you* tried selling your own stuff to yourself to get rich, people would laugh their asses off at you for being so foolish. But when a billionaire does the exact same thing, somehow he actually does get richer.

It's very easy: Just don't. The idea that Copilot or ChatGPT are suddenly essential tools is just buying into the corporate ad blitz. Games were made just fine (well, relatively) up through...what, 2023? 2024? They didn't suddenly become VITAL for budgets or design docs in the last 2-3 years.

@timnitgebru.bsky.social lays it out well here. For more from me and @alexhanna.bsky.social see thecon.ai and also our podcast Mystery AI Hype Theater 3000

That is not, in fact, a sex house. #Iceland #Reykjavik #photography #landscape #nature #naturephotography #landscapephotography #streetphotography #abandoned #decay #trees #graffiti

"I think refusing is actually the more hopeful, expansive vision of the future than the one that is telling us that the future is already settled and decided. That's my daring idea. Just say no."

if you're wondering why many people are not keen on LLMs, here is yet another excellent and entirely reasonable example. this is both a practical problem and also a PR and marketing problem for LLM proponents.

AI bros: perplexity is the best for research me: i used it for easily checked research and it failed in multiple ways AI bros: oh, not THAT kind of research. did you try google

There's also a phenomenon akin to Gell-Mann amnesia happening with AI where people easily recognize it's hallucinations and errors when it's discussing a topic they know about but they assume it's magically correct when discussing a topic in which they have no expertise.

imo, most AI skeptics nowadays actually have tried to use LLMs with some good faith and with the resources and training available to them, and are basing their skepticism on lived experience

i am actually working on an article right now about how the vast majority of the world's information - and so many current events - are not at all legible to LLMs, a fact that a startling number of people never seem to think about

It sometimes feels we're in a gish gallop (thank you en.wikipedia.org/wiki/Eugenie...) against people much worse than creationists (who are already annoying AF) because they really have caused statistics to somehow be seen as omnipotent to the point where it is allowed to destroy people and planet.

These two photos, were taken here in Iceland, 2018, during a trip to the remains of the WW2-era naval base in Hvalfjörður. The wall belonged to a military hospital #photos #iceland #photography

Here are three photos from Parc Jarry, spring, summer, and autumn. The photo of the bloodied gull is one of my personal favourites. The gull had tried to snack on a duckling and all of the duck mamas ganged up and it was lucky to get away with its life #montreal #photos #photography

I think the root of the “AI” evil is when AI researchers in the 1960s recognized that they outrageously underestimated the complexity of the human mind. They became humiliated by their promises that AGI was just a few years away—and then went full goblin mode that’s lasting to this day.

I am thinking that a surprising amount of good could be done simply by banning chat interfaces to LLMs. Most of the plausible use-cases that might turn into something good don't use them, basically all the awful ones do.

If AI were real, it would be using all the data companies already have on me as its primary input, and do stuff almost entirely on the backend. But the tech mostly relies on the ELIZA effect to appear impressive, and companies want the perceived markerability of visible "AI" in the product.

I’ve reached my Montréal years in my impromptu photography retrospective (2016-2019). 2019 was the year where, for a variety of reasons, I knew I wasn’t going to be living there for much longer. These are mostly from Parc Jarry and the Jean Talon market. #montreal #photos #photography

Spending the holidays processing just how betrayed I feel by the software and web industry My career has been fifteen years of hope and promise—some fulfilled, some not—followed by a decade where the industry stopped pretending and just looted society Being a bit angry about it is quite reasonable

It's like all of these hucksters and charlatans aren't actually technologists. You don't have to threaten people to adopt a technology. If you do, it's worthless. You don't have to say "OH NO. YOU'LL BE LEFT BEHIND!" It's all a con. People looking to make a ton of money and get out fast.

Creative industry has been maimed by technofascists and their marauding yes-men. My friends are losing formerly stable jobs. The AI bubble decimated the PC market. And indie storytelling is on life support because we are in a self-imposed recession. The Creator economy is dead. Shit is dire.