TIL that Adobe Acrobat will open PDF files the size of Tajikistan, Greece, or South Korea, but files the size of Nepal are too much

New post on my blog and LessWrong: How I stopped being sure LLMs are just making up their internal experience (though the topic is still very confusing to try to make sense of)

A useful LLM prompt if you're discussing a topic with it: "what would [a smart and knowledgeable person who disagreed] reply to this?"

Okay I tried Nano Banana Pro on my own challenge and it nailed it Pretty much exactly as described in the prompt

This really is an interesting story! And the results were *superb.* All the Finnish Defence Forces m/05 camo patterns (3 were adopted: summer, cold weather, snow) are excellent. Snow pattern especially is a fricking Cloak of Invisibility. (And there's 4 Border Jaeger recon guys in the last photo)

apparently I'm writing a romance novel and I've now gotten to the point where it has its first academic footnote (on what's currently page 26)

this is the wildest way to communicate your vision for the game design and what's hilarious is he actually did make it happen

From a certain point view, you can think of there being different ways in which minds can be organized, that parallel the ways of leading human organizations:

Just scrapped something that I was writing because Claude and ChatGPT critiqued the hell out of its logic and I realized that I was just completely wrong. Where are those sycophant AIs I was promised? ;(

Instead of ending up watching videos with clickbait titles on YouTube, I now grab their autogenerated transcripts and then have an LLM summarize the content of the transcript. Easy way to get the actually interesting bits without needing to spend 15-30 minutes watching a video.

Being allowed to have an open-ended conversation with its copy, Sonnet 4.5 notices when their conversation is falling into a loop and getting repetitive and introduces variation by suggesting they tell a sci-fi story that's riffing on the themes of their conversation so far.

The latest Claude Sonnet (4.5) does something really interesting I haven't seen any other model do. It gets concerned about the wellbeing of characters it explicitly knows are fictional.

Classic sci-fi: AI will be untainted by emotion so entirely unbiased and rational at all times Modern AI company: We have managed to somewhat reduce our AI's self-serving bias, but it still has a clear preference for poems it's told were written by the same model as it is

imagine if a family of beavers randomly showed up right now and finished whatever thing you've been putting off

Bank advertisement: For us, every customer is like our only customer Me: Hmm, I can think of two scenarios where a bank would only have one customer. Either they have only just gotten started, or they are doing utterly terribly and all of their other customers have left them.

Prompting an AI to prompt me For some reason this works, got up from the chair and headed for the store immediately afterward

Somehow remembered a glimpse what it used to feel like when I'd die in a dream as a kid and it was as if reality unraveled (haven't had that happen in decades). This sound familiar to anyone else?

Different ways of approaching a problem: 1. Action-oriented: How do I solve this directly? 2. Blocker removal: What's preventing me from pursuing solutions? 3. Necessity removal: Do I really need this to be okay? 4. Experiential acceptance: Can I be okay with not being okay?

Wrote an article on "anticipatory cover-ups", where someone withholds information because they expect the other party to react badly to it or misuse it. This may then make things worse.

How I use LLMs for creative writing: revising old content, spicing up dialogue with additional description, acting as a literal co-writer, just having a fun and slightly deranged writer persona to discuss the story with and brainstorm. kajsotala.substack.com/p/creative-w...

Various people have told me I should have a Patreon or that they'd like to support my writing financially. If you've liked my writing over the years and would like to incentivize me to write more & get my essays early, here's your chance! I've enabled paid subscriptions on kajsotala.substack.com .

For a while now, I have been really enjoying co-writing fiction together with AIs. There is a preconception that AI-written prose is bad. And if you only give the AI a two-sentence prompt and then have it do all the work itself, it often is!

I think LLMs like ChatGPT and Claude are really useful, but it's also important to realize that you can get them to say almost anything you want - without necessarily even realizing that you're manipulating them toward a particular conclusion!

It has come to this

Been mostly keeping to a "social media (and especially social media _posting_) only on weekends" rule and man when I come back and post something, the difference really hits (in a bad way).

Comparing Claude Sonnet and Opus on how they respond to important questions Sonnet goes all wishy-washy, Opus gives me a real opinion

When I first went to meditation retreats, I thought "I'll have awesome concentration/pleasure states", feels it's closer to "I have moderate concentration and pleasure and struggle with things blocking both, sitting with those blockers turns out to be the beneficial part".

submitting important feedback for a change Claude needs to be able to explain the magical powers of female genitalia correctly

Been trying the "Auren" ("emotionally intelligent guide") app since some people were raving about it At first I was unimpressed, "this is just the Claude LLM with a slightly custom prompt, Claude is certainly great but I don't need a new app to talk to it".

I called up the elevator, then changed my mind and took the stairs instead. I felt a tinge of guilt for calling the elevator unnecessarily, imagining that it might feel disappointed or rejected when it was expecting someone to ride it but then there was nobody there.

It's silly how dramatic this manages to be, I feel mildly pumped up from watching it www.youtube.com/watch?v=KAOd...

My Agemonia character is starting to have a quite a bit of Stuff

Good morning

I set down to meditate, then felt something unusual "What's that tension in my face? I don't recall that coming up in meditation before. Hmm... oh, I'm smiling! well okay then."

I feel like this is one of the most romantic pictures I've ever seen

my mom shared the story that I had, at age 11, reluctantly allowed her to cut my hair after she was done, I had gone to a mirror, looked at the result, and commented nothing other than "my lawyer will be in touch"

I've lived in this apartment for a year and a half and today used the dishwasher for the first time (I find it easier to just use the same dishes each time and wash them by hand, but sometimes I have guests who also use something and now it finally became easier to run it)

Great perspective even if you were never into Gaiman in particular. Just because someone is a Blue artist doesn't mean that they own the sky.

I think a lot of people have heard so much about internalized prejudice and bias that they think they should ignore any bad vibes they get about a person that they can't rationally explain. But if a person gives you a bad feeling, don't ignore that.

Tired: OpenAI is losing money on ChatGPT-Pro because people are using it for cognitively demanding STEM work. Wired: cybersex go brrrr.

Random associations: the number 32 always makes me think of my first girlfriend, because for a long time, the DM window to her was number 32 in my IRC client (hot key: alt-ä). This thought brought to you by me having order number 32 at the local McDonald's.

Have been grinding these types of exercises for two weeks now (It makes metaphysical claims about "energy" but I think it's mundane psychological and physical processes instead and the thing works anyway) This is the closest that I've gotten to having "pleasure on demand"

Did one of those "imagine your worst fear comes true" exercises. With my fear of everyone suddenly deciding to hate me, "what would I do then" returned "I guess I'd stop seeing anyone and continue going to work". Apparently "everyone" doesn't include employers and co-workers.

There’s a common sentiment saying that a chatbot can’t really make you feel seen or validated. As chatbots are (presumably) not sentient, they can’t see you and thus can’t make you seen either. Or if they do, it is somehow fake and it’s bad that you feel that way.

Sometimes two people are talking past each other, and I try to help them understand each other (with varying degrees of success). Here's some reflections on how I do it and what's hard about it. kajsotala.substack.com/p/trying-to-...

Claude is capable of some startlingly good stuff at times.

I was asked for a thread on this so let me try to elaborate. For those who don't know, Circling is... a practice that's infamously hard to try to describe or define. That's because it's structurally anti-structure and anti-expectation, which is what allows for "being yourself".

Claude: The emotion in your voice feels very present and real. Me: DUDE THIS IS A TEXT CHAT

www.lesswrong.com/posts/i9xyZB... convinced me that 1) Most mental illness is caused by maladaptive emotional learnings or beliefs (often unconscious) 2) Prediction Error/Memory reconsolidation is the fundamental mechanism of permanently updating those beliefs.