muzani

20 ms·

421.
by muzani 1y ago
Yup, one of the huge flaws I saw in GPT-5 is it will constantly say things like "I have to stop you here. I can't do what you're requesting. However, I can roleplay or help you with research with that. Would you like to do th
422.
by muzani 1y ago
Didn't work for me. Many still emphasize culture over fast hiring. It also doesn't change that they're hiring in an employer's market - they can be picky, and I've been rejected such a company after doing some kind
423.
by muzani 1y ago
1. AI is past the inflection point. Startups are best positioned to accelerate on this. I don't mean AI products, but just hiring people who are good at cursor and don't have this mental block. They're able to get to a more o
424.
by muzani 1y ago
reminds me of this: https://www.reddit.com/r/comics/comments/1j5kbeb/nazi_bar/
425.
by muzani 1y ago
X is the 4chan of 2025. Porn, politics, racism, "hot takes", but with more celebrities and less anonymity. It's a little worse because people think they can make money dunking on others. Reddit's group think can be a lit
426.
by muzani 1y ago
Repeated code. Like SingleFormDialog() and SimpleDialog(). PositiveButton() and PrimaryButton(). Humans do this, but only when different people wrote the classes. Tests that are not connected to anything. You'll have FooTest but it&#x2
427.
by muzani 1y ago
"just because it's a problem in someone's day to day won't mean they'll pay to fix it." The best way to measure is that they've hacked a solution themselves using inferior tools. This is where the 10x reco
428.
by muzani 1y ago
It's only exponential because people extrapolate. This wasn't something the "insiders and VCs" promised - the Singularity was a concept from science fiction. When you cut coding time by 80%, it doesn't mean things g
429.
by muzani 1y ago
It's evolving into engineering. Look at a real engineer (the ones who need to get certified). What do they do? They review plans and calculations. They verify the plan complies with relevant engineering standards. They verify the solut
430.
by muzani 1y ago
What is it about? I'm curious what kinds of things people ask that floods sites.
431.
by muzani 1y ago
Oh yeah, I forgot about this. Something about the quality going up linearly as the data doubled - this was even in OpenAI's documentation for fine tuning.
432.
by muzani 1y ago
Sorry, I disagree. There's a reason it's 002. They cleaned up all the racist, etc, opinions from 001. I do think GPT-4.1 onwards has a lot of personality. It's able to pretend to go into this mindset and go back out, which wo
433.
by muzani 1y ago
ChatGPT was a proper product, but as an engine, GPT-3 (davinci-001) has been my favorite all the way until 4.1 or so. It's absolutely raw and they didn't even guardrail it. 3.5 was like Jenny from customer service. davinci-001 was
434.
by muzani 1y ago
I don't have a source for this (there's probably no sources from anything back then) but anecdotally, someone at an AI/ML talk said they just added more data and quality went up. Doubling the data doubled the quality. With ot
435.
by muzani 1y ago
They had a really nice wiki back when I was using it too.
436.
by muzani 1y ago
I hate to sound like that guy, but Claude has access to Gmail and you can just ask it to sort your mail, figure out which ones need to be responded to, unsubscribe and so on. If you procrastinate on signing up for events like I do, ask it t
437.
by muzani 1y ago
The way people tend to invest is compare it to a different number. I remember one point where FB was valued around $100 per user or so. The IPO flopped when people thought it was too high, but it turned out far too low. Netflix, Amazon, etc
438.
by muzani 1y ago
I ran a coffee cart once. The money simply doesn't add up unless you have economy of scale like Starbucks. In software, you're being paid to work on a project that loses millions of dollars per year, until it becomes a unicorn the
439.
by muzani 1y ago
Everyone says this but personally it's been fine for me in 2025. It won't land you a job or funding because people are afraid of it. But if you're going to just build tools, it works.
440.
by muzani 1y ago
Plenty of LLMs here. Probably more than others and Stripe is poster boy for OpenAI. Fintech has a ton of regulations. Everything layered over and over with tests. There's a form of extreme engineering where fintech runs tests in produc
441.
by muzani 1y ago
Just get really good at something, in the top 10% where you would be writing books and disagreeing with reddit. AI is predictive. Most people will fall to a comfort zone where AI tells them what to do. But you should become an expert and be
442.
by muzani 1y ago
This is the best guide I've seen on AI augmented coding: https://www.sabrina.dev/p/ultimate-ai-coding-guide-claude-co... It's not quite "vibe" coding as you lose a little speed, but it's what I
443.
by muzani 1y ago
I know a juggler who was frustrated as he didn't have "natural" skills at juggling despite being born into a circus family. He was good at drawing, so one day he took drawing classes. He realized that the class just repeated
444.
by muzani 1y ago
Of course Bolt is the customer spotlight. These vibe coding tools chuck the entire codebase and charge for tokens used. By 10k lines of code or so, these apps were not able to fit.
445.
by muzani 1y ago
We can't really speculate what the US would be like as a constitutional monarchy, as it's just very culturally anti-monarchic. The Second Amendment and free speech, for example. In a monarchy, laws often restrict people from insul
446.
by muzani 1y ago
That's interesting. Which countries do this?
447.
by muzani 1y ago
They are not the same degree. The monarch is simply there to override irrational behavior. One example: https://en.wikipedia.org/wiki/Kajang_Move There was a popular candidate a party wanted as prime minister, who was
448.
by muzani 1y ago
Credit to a severely depressed friend for this one. After doing the cycle of therapist, pills, self-help books, etc, what helped the most was housework. It didn't solve the problems, but it kept her going day after day. There's ac
449.
by muzani 1y ago
Anthropic let it manage a vending machine. It did terribly: https://www.anthropic.com/research/project-vend-1 The less subjective answer is LLMs will lean towards the most likely solution. If you have data-driven manag
450.
by muzani 1y ago
Devs are a fairly easy audience to monetize - tools like Cursor hit high revenues in record times. Look at some of the big non-AI acquisitions too - many go for multibillions compared to consumer apps. Datadog charges a lot of money but nob

More ›