muzani

31 ms·

721.
by muzani 1y ago
It's trained on the Google style I guess. Google code always feels excessively commented, to the point where I delete comments from Google samples so I can read the code.
722.
by muzani 1y ago
I planned to do this one. Check with the law, like talk to a lawyer or someone in the consumer ministry. Prices were protected under law in our case. It wasn't legal to crowdsource it; stores had to share it willingly. I'm not sur
723.
by muzani 1y ago
What is the alternative though? Do I boycott it? Buy my games from Epic Store, book taxis while I'm on the train, ignore messages from teachers, and do all my banking from ATMs? Apps are still more convenient and cheaper. Admittedly di
724.
by muzani 2y ago
It's this app: https://play.google.com/store/apps/details?id=com.tongtongsu... It's not the same company as the one that made the door. To be fair, the door is a one time purchase and they shouldn't
725.
by muzani 2y ago
Unfortunately, buying basic things requires an app now. Paying off my credit card and house requires an app. Getting a "taxi" needs to be done by app. School updates for tomorrow's homework go through Telegram, telling the te
726.
by muzani 2y ago
I do both lol. Flutter is fun to play with and vibes very well with AI. But as a day job, things like dependencies, animations, design systems, and super customized things like Xiaomi are the bane of my day.
727.
by muzani 2y ago
Well I have an old laptop that can't upgrade to W11. The battery is dead and bloated. It overheats when playing Roblox. It doesn't even handle Marvel Rivals. The camera and speaker no longer work. Two keys have peeled off. This mi
728.
by muzani 2y ago
OpenAI has always been somewhat controlling of their output. They used to require that anything that uses their API go through a verification process to prevent it from being used for fake reviews, astroturfing, etc. This was the main criti
729.
by muzani 2y ago
There's projects like Cara which are vehemently the anti-AI Instagram. Steam also blocks AI projects unless you can prove the AI images were ethically trained. Itch.io is a bit softer and requires a declaration of AI use so some groups
730.
by muzani 2y ago
How do we detect it? Most of it looks like entropy. You can justify that the phone is unrepairable because it was necessary to pack everything tightly to make it thin. Once there's a way to measure it, we can discuss what if.
731.
by muzani 2y ago
I would actually like to see something like TV Tropes in spreadsheet form. Something like this: https://tvtropes.org/pmwiki/pmwiki.php/JustForFun/OverdosedT... But in more detail like category etc and coverin
732.
by muzani 2y ago
Unlikely. We use AI extensively now - it shortens parts of the schedule, but you still need a "programmer" to connect the endpoints. SaaS didn't take jobs. Frameworks didn't take jobs. AI will take quite a few, but that&
733.
by muzani 2y ago
Ads worked back in the era where people couldn't make payments online. These days it's just more effective to have them pay, or in some cases, pay to disable the ad.
734.
by muzani 2y ago
Strong yes. AI is the ultimate generalist, but it's not a specialist unless the operator is a specialist. It'll output the average of what it's trained on. It may know the new stuff, but unless someone programs it to return
735.
by muzani 2y ago
I've been using something like Google Family Link, which works fine, except that it ties in to Google Family, along with YouTube, Play Store, Google One. I'd have to kick my sister out of the group to monitor another daughter and
736.
by muzani 2y ago
Yes, please. I've been looking for something like this. Lately I've been just casually going into another language with ChatGPT and asking it to correct me. I do I like some of the old languages, things like Aramaic, which just ha
737.
by muzani 2y ago
The data isn't there. Most would be out of date, and there's no incentive to train a LLM on this. OpenAI doesn't even do a lot of things well; their RAG, STT, TTS, voice, embeddings is below average, but it's convenient
738.
by muzani 2y ago
It's going to be more about engineering . We don't need to know how transistors work to make a website. But moving forward, you do want to know patterns like repositories. You want to know terms like "decompose a conditional
739.
by muzani 2y ago
Please don't put too much pressure on it. This is how it turns us into paperclips. Also studies have shown that under a lot of pressure, they actually perform worse.
740.
by muzani 2y ago
You need a crowbar. Enter at an angle, then pry it off. Slack is my favorite example. It looked like chat. It had a hash, like IRC channels did. But it was actually email! It cut down back and forth in email. You can just give someone an em
741.
by muzani 2y ago
The quoted paper is from 2018. Evolutionary programming is IMO similar to a search algorithm. The biosphere is not. It's not searching; it's removing unfit matches. Which is far less efficient. I'm a big fan of evolutionary p
742.
by muzani 2y ago
We ask this question in interviews too. One purpose is we want people who disagree and can handle this maturely. Everyone has negative experiences. People who don't have likely never tried anything difficult. But you have to be diploma
743.
by muzani 2y ago
My general advice is to be aware when you're in the grief cycle and get out of there as fast as possible so that you can think clearly. A lot of people in the comments are in denial and bargaining. You sound like you're at depress
744.
by muzani 2y ago
Engineering is being good at using engines. Knowing all the systems that this engine is composed of, optimizing it, making it do what it's meant to do with minimal error, coming up with tests to verify that it does, facilitating all te
745.
by muzani 2y ago
Cursor is the popular one now for good reason. It does have gemini-2.5-flash-exp, along with all the claudes, openAI, deepseek, and their own free ones. It comes down to about $16/month, which includes limited use of claude sonnet and
746.
by muzani 2y ago
Ah, the anime variant.
747.
by muzani 2y ago
I expect some manual correction after the work is done. I actually mentally counted all the times I pressed backspace while writing this paragraph, and it comes down to 45. I'm not counting the next paragraph or changing the number. Hu
748.
by muzani 2y ago
I saw one on LI with an image. It ended with a ChatGPT style question like, "Would you like me to explain X in more detail?" Nobody noticed. It seems they only see the image and read the first paragraph.
749.
by muzani 2y ago
The real news for me is GPT 4.5 being deprecated and the creativity is being brought to "future models" and not 4.1. 4.5 was okay in many ways but it was absolutely a genius in production for creative writing. 4o writes like a ski
750.
by muzani 2y ago
One of the major advantages and disadvantages of LLMs is they act a bit more like humans. I feel like most "prompt advice" out there is very similar to how you would teach a person as well. Teachers and parents have some advantage

More ›