HN user
Oh, I'm glad I don't work in the oven business. We're just starting a stealth startup that's revolutionizing dishwashers, and the prototypes are amazing. They use less water, less detergent, and this weekend we're hoping to solve the last remaining issue: occasionally, they break glasses.
So you're saying I can absolutely have a deep, nuanced discussion with an LLM, as long as I don't ask how he arrived at his conclusions?
You say you can have increasingly nuanced discussions with stronger models.
What I say is, when I asked Claude why he applied a certain change I didn't understand, and boy, it was a small change, he said he "reasoned from first principles" based on the code paths. But it didn't work, and when I asked, "Okay, describe the steps of your reasoning from first principles," it literally answered that it had just made it up.
So, nuanced discussions with models, I don't buy it.
Is it really cheating if you work hard and memorize all the questions? That's basically learning.
Just 300 a day? That's only one ticket every 1.5 minutes. I hope in a year we can fix an issue under 30 seconds with ZERO supervision.
Why does the AI need SQL queries? Who needs that? It will just write its own ACID-compliant database with its own language, and while it's at it, reinvent the operating system as well. It's turtles all the way down.
To continue, this design has additional benefits:
The code is not littered with dry-run flag checks; the internal code doesn’t even know that a dry run is possible. Everything is rolled back at the end if needed.
All database referential integrity checks run correctly.
Some drawbacks: any audit logging should run in a separate transaction if you want to log dry runs.
I use a similar strategy for API design. Every API call is wrapped in a large database transaction, and I either roll back or commit the transaction based on dry-run or wet-run flags. This works well as long as you don’t need to touch the file system. I even wrap emails this way—emails are first written to a database queue, and an external process picks them up every few seconds.
love the title of Windows recovery process screen: uninstall latest quality update
could you share an anonymized desk photo?
I've just got a new Samsung and it's wobbling too. I hate this. Why can't they at least put the cameras in the middle? Or maybe horizontally centred? Or they could just put another bumper on the other side to make it symmetrical. I'm looking for a cover to balance this out, but no luck so far.
When I started high school in the early '90s, there was a compulsory summer reading list of 10–12 books, each ranging from 300–800 pages. Then we had to write essays about them. This was just our summer homework before the new school year started. I didn't enjoy it at all; at the same time, I read lots of easy fiction, sometimes several hundred pages a day.
My six year old (who is still in kindergarten) reads about 70–100 pages per week of books aimed at eight to nine year olds.
What about reusing a CTE? Let me import a CTE definition so that it can be used throughout my app, not just in the current context.
Just two days ago, I asked ChatGPT to provide an explanation of the place-value system that my six-year-old could understand. The only problem was that it mixed up digit value and place value, which caused it to become confused. I spotted the mistake, and ChatGPT apologised, as it usually does. But if my six-year-old had asked it first, she wouldn't have noticed.
I'm not sure how much misinformation my child would learn as truth from this device.
For me, it seems the issue with the recipe was due to pre-recorded segments. The guy quickly realized that the "AI" was ahead of the script and cut it short to avoid making a bigger mess during the demo.
As for the Zuck segment, I believe it was a race condition.
Here is the video:
https://www.reddit.com/r/interestingasfuck/comments/1nkbqyk/...
The last time I asked for a code review from AI was last week. It added (hallucinated) some extra lines to the code and then marked them as buggy. Yes, it beats humans at coding — great!
A PhD-level degree in fabrication.
The last time I asked an LLM AI to review my code, it added extra lines that weren't originally there. Then, it claimed that there were many bugs in those lines. When I pointed this out, it apologized. But I don't know how much time I wasted on this thing. I'm a developer with over 30 years of experience. Should I ease into vibe coding?
I have two iPad Airs, first generation, but the latest YouTube app no longer works on them. Skype and Teams don't work on old iOS either. Do you have any ideas other than using them as photo frames?
and then there is Marco Pierre White, How to finely chop onions:
But you can click on the chicken and it will jump. KPIs matter.
I taught my five-year-old to read in approximately three months. We progressed from simple letters to sounding out two letters, then three, then words with hyphens, then simple stories, and finally children's books. We practiced daily for 5-20 minutes, five times per week. There were plateaus, but also huge jumps in ability, especially when we took a two-week break. Now, she can read two to three pages of an A4 document in one sitting with great speed (i.e., she looks at the words and sentences in one quick pass). She still slows down and makes some errors with complex new words.
Note that our language is Hungarian, which is much easier to teach because writing and sounding out words are nearly one-to-one in terms of letters and sounds. The AI part: Phonemic orthography: A writing system in which each letter (or combination of letters) consistently represents a specific sound (phoneme), and each sound is represented by a consistent letter. Hungarian is highly phonemic, meaning you can usually tell how to pronounce a word just by looking at how it's written, and vice versa.
I'm just wondering, if AI replaces programmers, will our super AI also fix all the slowness and bugs in current software?
I saw an Anthropic Claude guy talk about how AI will replace most programmers within 2-3 years, but definitely by the end of this decade. I suppose an additional data center of AI agents could also fix all known bugs.
Yesterday, I asked AI for help:
Check my SQL stored procedure for possible logical errors. It found a join error that I didn't remember including in my SQL. After double-checking, I found that it had hallucinated a join that wasn't there before and reported it as a bug. After I asked for more information, it apologized for adding that.
I also asked for a C# code with some RegEx. It compiled, but it didn't work; it replaced the order of two string parameters. I had to copy and paste it back to show why it didn't work, and then it realized that it had changed the order of the parameters.
I asked for a command-line option to zip files in a certain way. It hallucinated a nonexistent option that would be crucial. In the end, it turned out that it was not possible to zip the files the way I wanted.
My manager plans to open our Git repository for AI code and pull request (PR) review. I already anticipate the pain of reviewing nonsensical bug reports.
A totally broken function that a single person uses once a month to save five minutes isn't nearly as valuable as a well-designed feature that 5,000 users rely on daily to save just one minute each session.
We shipped our VB6 software on a 1.44 MB floppy disk. Actually, we had to add some random stuff to the installer to make our software fit on two floppies instead of one, so that customers would think our application was more complex. Our CEO wanted to make sure that the end user had to eject disk 1 and insert disk 2 during the installation. Sometimes you want to go big. So size matters.
The question is, can you guarantee my safety?
I'm wondering if you feed all the Google street map photos into a special ML designed just for that, how important could that be for say the CIA or FBI?