sixtram

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?

I use a 4 liter home distiller. Without specific brand just google "home water distiller 4l".

We use an ultrasonic humidifier all the time. At the same time, we use only distilled water (we distill it at home every 2-3 days).

First, this removes minerals from the water, so the room isn't covered in "white dust" after using the humidifier. But even with this, biofilm can build up inside the machine, which is not visible, but you can touch it. I also remember that when I used just plain tap water, the biofilm built up faster.

This article definitely hit me, I realized that we need to clean the humidifier more often.

I used Mithril in 2015 for a very complex form view in an otherwise vanilla jquery page. It saved the project. React was also a new player at the time, but for some reason I wasn't able to include it with the "require.js" bundler (I guess the module format was not standardized yet). I actually liked Mithril, sent them some bug reports.

Well, I don't know, I live in Hungary and just this morning I checked what's on the menu at my child's kindergarten. For breakfast they have margarine with honey on white bread, I'm not sure if that's good. For afternoon snack they got "kakaóscsiga" which is a pastry full of chocolate and sugar.

Turned out it's a small Hungarian company at Budapest. Their website went silent this morning.

Thanks for the help, you were right.

Does anyone know what tool and font was used in the illustrations in the article?

And did you fix the performance problem?

Great link, I'm going to add this to my list of favorite interview questions. (^_-)

I also like the magic link emails. I'm actually a developer on a SaaS product that makes heavy use of magic links. One downside - other than security - is that email delivery is not instant and these emails can end up in the spam folder. The wait can be frustrating. Magic links are usually good for users who are invited into a platform, for example to review a shared document or something.

Welcome home, my friend. Hungary always sucks, but at least now we are also last in consumption in the EU, time to catch up with Bulgaria.

I always liked Jim Gray's marbles example to demonstrate the problem with snapshot isolation, so much easy to understand:

https://techcommunity.microsoft.com/t5/sql-server-blog/seria...

Is this a native app or Electron stuff?

To derive the payload capacity of a self-landing ballistic rocket, we need to consider several factors and steps in the analysis. Let's break it down step by step:

Step 1: Define the Requirements We need to establish the basic requirements and factors that influence the payload capacity:

Rocket Parameters:

Total mass of the rocket (M_total) Mass of the rocket without payload (M_empty) Mass of fuel (propellant) required for landing (M_fuel) Specific Impulse of the rocket (Isp)

Payload:

Desired payload mass (M_payload) Rocket Equation:

The Rocket Equation will be fundamental to our analysis:

............... ............. ........... ........... .................. ............. ............

So, with the given parameters, the payload capacity of the self-landing ballistic rocket is approximately 9,790 kg. This is the maximum mass of payload it can carry and still be able to perform a controlled landing.

Stonebreaker is 80 years old and he makes a new company. I'm almost 50 and wondering when my tech career will end. I would be happy if I could start a new company at 80.

A recent Stonebreaker video:

https://www.youtube.com/watch?v=eB4bJqDzsU8

>> As someone with a top quintile-level of mental toughness and grit

Do you have any resources or recommendations on how to achieve a high level of mental toughness and grit?

you ssh-d into 25K servers one by one? I mean, manually?

true, at the same time your bumblebeelabs site is down ...

Is he on hormone therapy? I guess when you're a billionaire the doctors mix you up a good cocktail.

Maybe because Paul Grahams wife Jessica Livingston is one of the financial backers of OpenAI?

A google search returns about 10,200 results for this, gpt4 probably picked up this sequence as well during the training.

I tried some basic math and algo questions with both GPT-3.5 and GPT-4. I'm impressed how it can spit out the algorithm in words (obviously because of the pre-training data), and how it then can't follow with the algorithm itself. For example, converting really large integer numbers to hexadecimal. Or comparing two big integers, it starts hallucinating numbers into it. It may be able to solve an SAT exam with a high score, but it seems you can pass an SAT exam even if you cannot compare two numbers.

He has huge problems with lists or counting. If you know more or less how LLMs work, it's not that difficult to formulate questions where it will start making mistakes, because in reality it can't run the algorithms, even if it spits out that it will.

I don't know. OpenAI claims that GPT-4 can be in the 89th percentile on the SAT Math exam. At the same time, I asked GPT-4 to compare two six digit number and in some cases failed. When I asked for a why, it started to make up completely false math rules, e.g. to compare six digit numbers you need to check their 7th digit, which is not just nonsense (where is the 7th digit in a 6 digit number) but shows a complete lack of understanding.

So while I believe it can produce "accurate" answer for SAT questions, as it was trained on those, I also believe that it doesn't have any kind of math logic it follows.

I'm not sure this dichotomy can be solved with "more" training. It's also telling that in some cases it fails, and in some cases it doesn't, you just need to change the numbers, even with simple numbers.

Why do we stop with hyphen, n and m dash? There are at least 30 different use cases, we should not reuse only 3 versions of some short line. Let's make 30 versions, one for each meaning. (cynicism)