Timo Mihaljov

Timo Mihaljov

@timo.mihaljov.com

0.1x programmer

408 followers1K following495 posts

Timo Mihaljov reposted

a lot of people have encountered the overconfident stem nerd, the guy who programs and so everything else must be easy from the outside it does look like "i'm smart at code, so therefore i can do anything", but i promise you that they're like this with code too

Timo Mihaljov reposted

Replying to @hellgorithm.website

AI enables developers to think less about implementation details and focus more on what they actually care about: nothing. Nothing at all.

Timo Mihaljov reposted

This is exactly the principle that has fucked up everything from deli flyers to UX in billion-dollar products Once upon a time, even if you were doing something on your own, it took long enough to do that it made you think about what you were doing! You had time to think "hang on, this sucks."

I think the thing about AI and small business owners is that previously the design would have to be run past either a second person who know some amount of design, or they'd learn photoshop and some amount of design with it now they can ask the AI to do their bad design and the AI never says no

Timo Mihaljov reposted

The first genuinely honest ad for generative AI I've ever seen, from Microsoft / CoPilot A mediocre white guy sees his non-white colleague getting praised for good work, he asks AI to *literally copy it*, and then he gets all the praise instead: www.ispot.tv/ad/SoDi/micr...

Timo Mihaljov reposted

Replying to @philbybear.bsky.social

"Every respectable artist in this country was against the war. It was like a laser beam. We were all aimed in the same direction. The power of this weapon turned out to be that of a custard pie dropped from a stepladder six feet high." --Kurt Vonnegut

Timo Mihaljov reposted

This is it. This is my breaking point. We have to stop. We have to stop with computers. We have to stop computers. We need to stop computers so we can take a break for a decade and then build them back better.

Timo Mihaljov reposted

people were 100% convinced ELIZA was conscious even when you could see its 200-some lines of code and even when the guy who wrote it yelled at them over their foolishness

Mezentine@mezentine.bsky.social

One must ask which is more likely: that we have developed genuine, self-aware machine intelligence, or that a piece of text generation software that's been hand-tuned for thousands of hours to portray an intelligent conversationalist is affectively effective on a large number of people?

Timo Mihaljov reposted

i've been saying "knowledge transfer is dead, people don't ask their coworkers questions anymore" for a while now, but honestly it's kinda funny that we're setting out to build a god and ended up with a tower of babel

Timo Mihaljov reposted

I would simply choose a measure that remains a good measure even after being made a target

Timo Mihaljov reposted

In the vein of "public libraries could never be created today," a pitch for a calculator would get you thrown out of the boardroom. What do you mean, it just maths? Where are the ads supposed to go? How does it send usage telemetry back if it doesn't even have wifi? There isn't even a leaderboard.

This post is unavailable.

Timo Mihaljov reposted

Coding agents reduce software development to a simple matter of: * specifying requirements * verifying correctness and quality * maintaining and operating software you didn't write As we all know, those have always been the easiest, most well-understood parts of the craft, so everything is fine!

Timo Mihaljov reposted

Tekoäly-, kielimalli- ja datakeskussekoilu saa toden teolla pohtimaan sitä, mistä ihmisyydessä lopulta on kyse. Tämä teknovelihybris on nimittäin täydellisesti vastoin omaa ymmärrystäni ihmisyydestä ja inhimillisyydestä. Oikeastaan se näyttäytyy minulle jonkinlaisena antihumanismina. #yhteiskunta

Timo Mihaljov reposted

My grand unified theory of this moment, after a few months of intense and variously high-stakes engagement with various corporate and public systems, is that pretty much nothing works. We just have each other.

Timo Mihaljov reposted

one thing you learn from making a lot of things is that the nitty gritty details, the millions of tiny invisible decisions, ARE the art. The "higher level ideas" are not nearly as important or interesting

Freya Holmér@freya.bsky.social

"AI lets you focus on the higher level ideas!!" because we all grew up dreaming of becoming managers and CEOs we all know the real artists at the very highest level are the shareholders, famously full of amazing ideas

Imagine thinking "There's not enough software in the world, and what little we have is made with too much damn care! We don't need this kind of perfection – we need more apps!" I'm utterly unable to relate to the people in my field.

AI coding tools are no different from compilers: you take the compiler output, make your coworkers review it, and commit it to source control so other people can use their own compilers to edit it, just like we've always done.

Timo Mihaljov reposted

A thing I learned about Meta when I was trying to understand how their statements relate to their actions a few years ago is that *they just tell absolutely indefensible lies* and journalists almost always let them. They believed no one would ever punish them for it in any way that matters.

Justin Hendrix@justinhendrix.bsky.social

Most state attorneys general released Meta from Cambridge Analytica-related claims as part of the child safety settlement. New Mexico wasn't part of that deal, and a jury there just found 26 of 29 past claims about Cambridge Analytica, hate speech and misinformation willfully deceptive.

> Goldman Sachs warned Friday that AI hyperscalers, including Meta, need to generate roughly $300 billion a year in AI services revenue just to break even on their capital spending—and $1 trillion annually to earn meaningful profits—spooking investors. finance.yahoo.com/markets/stoc...

finance.yahoo.comMark Zuckerberg Loses $9 Billion In A Day As Goldman Sachs Pours Cold Water On Meta Stock RallyZuckerberg’s $9 billion loss was the biggest single-day drop among any billionaire Friday.

Some big tech 2025 revenues for scale: Amazon: $638 B Apple: $416 B Alphabet (Google): $350 B Microsoft: $282 B The AI industry would have to generate way more revenue than all of Microsoft, about the same as Google/Apple, or half of Amazon *just to break even*! That's a lot of Claude Max plans.

A chatbot is like a bucket of gasoline and a pack of matches when what we need is an internal combustion engine.

Glyph@glyph.im

New Blog: What Would A Serious AI Product Look Like? Every “AI” tool is missing critical features that you would need if you wanted to do real work with them. What would those look like, if they existed? https://blog.glyph.im/2026/09/serious-ai-product.html

I don't know if the company or the people behind Jev are as horrible as all the other players in the AI space, but I think they have the right idea about how to constrain LLMs to work as serious tools instead of a parlor trick that's got way out of hand.

Timo Mihaljov reposted

I wondered if I'd be excited about "AI" if I was just growing up now. Then I remembered that in my teens I used to spell Microsoft with a dollar sign and ran Linux back when the easiest way to get it was for some weirdo to mail you a stack of CD-Rs. So yeah, I'm pretty sure I would still be a hater.

Researcher: In my paper I discuss the ways in which automation may expand rather than eliminate problems with the human operator Tech Company: At long last, we have created the Automation from classic research paper Ironies of Automation

These days I often end up deleting my own code comments that explain why the code is written the way it is. They're shaped like Claude's nonsense, so no human will ever read them, but without any comments the code itself might look human enough to make someone actually read it.

This post is unavailable.

Timo Mihaljov reposted