ako

6 ms·

1.
by ako 3d ago
I’ll agree with your last 3 words: they’re broadly similar systems - systems of deceit and self-deception. They may share the same god, but for 99% they’re different enough that they think they can kill one another and their god will approv
2.
by ako 3d ago
Exactly, and all Jews think Jesus is a fake. And on top of that all the past gods, Roman, Greek, Inca, Vikings, etc.
3.
by ako 4d ago
There can be more than one solution.
4.
by ako 4d ago
You forgot “we can stop climate change”…
5.
by ako 4d ago
Come on, it’s way more common than that. We’ve invented 3000+ gods and almost as many religions, most of them are incompatible with each other. So, most of these must be incorrect, so a huge amount of self-deception. But as Harari argued in
6.
by ako 6d ago
There are plenty people that just turn on the radio or Spotify to have some random Muzak in the background.
7.
by ako 7d ago
No, for that money I can buy an iPhone and an iPad. Just checked, iPhone 17 256gb plus an iPad 13inch 256gb is about 100 euros cheaper than the cheapest iPhone duo.
8.
by ako 11d ago
Quality of life will be dramatically lower for many because we did not take action
9.
by ako 22d ago
Why does it matter? It’s like a named operation, and if you want local scope you can create CTEs, which is like an anonymous local operation. So you have both options.
10.
by ako 23d ago
Writing speed is not really relevant anymore, LLMs do it. Reading is what matters most.
11.
by ako 23d ago
1. Can be done using views. Creating named views is really not that different than building named functions or methods.
12.
by ako 23d ago
I thought you were supposed to eat the pudding?
13.
by ako 26d ago
Flipped classroom method?
14.
by ako 27d ago
This is not counter collaboration, collaboration can also benefit from the use of AI.
15.
by ako 27d ago
I see it differently: if you’re not willing to consider how you could be better/faster/more productive with AI, you’re unwilling to adapt to the future, and not flexible enough. You’re going to slow down your employer. Most busine
16.
by ako 27d ago
As I said the response of AI is very much based on the input and context i give it. It’s not like everyone gets exactly the same answers from AI. Ai is a tool that I control. Also, if the response of AI is sufficient, what is the added valu
17.
by ako 28d ago
If the response of AI is sufficient, why would i retype it in my own words? Just a waste of time. Most of what i have AI generate is based on my words and input anyways. I'll often write a draft note, and then have AI generate multiple
18.
by ako 1mo ago
Why are you so distracted by style? I see it too, it could be improved, it might be improved, but I don’t care as long as it get shit done. And it does, tons of shit gets done. And the communication is usually more informative than how cowo
19.
by ako 1mo ago
Exactly, the German autobahn is also awesome. /s
20.
by ako 1mo ago
Kitesurfing is so last decade, this decade is all about wingfoiling and parawinging.
21.
by ako 1mo ago
Pagerank is just a popularity contest, doesn't say anything about trust. AI probably does a better job determining trustworthyness than pagerank.
22.
by ako 1mo ago
Maybe, but it's reality. Not too long ago the rule would be "if it doesn't show up in google, it doesn't exist". Today agents+LLMs are better browsers than Chrome/Safari/Firefox/... If a webpage does
23.
by ako 1mo ago
The best way to read the information on the internet today is via a LLM. Just like you don't process raw data in a data warehouse or datalake by looking at tables, you use a SQL or BI tool, to process all the information on the interne
24.
by ako 2mo ago
Go value proposition for me, when developing with Claude code: 1) fast compilation times, faster iterations, 2) user friendly to install, one single binary per platform, 3) feels safer than JavaScript/typescript, which have too many np
25.
by ako 2mo ago
We’re way beyond that, LLMs are used for reasoning - plan, do, check, act. They will define a goal, decide how to get the info needed (run a script for example), and use these new insights to adjust the plan. This is way beyond just retriev
26.
by ako 2mo ago
Exactly, and a dsl gives you many options and tools to make the llm more reliable: a grammar enables deterministic syntax checker, error messages injecting correct syntax into the context, linting, terse examples in the skills for token opt
27.
by ako 2mo ago
For open source llms you can give your grammar to the llm to help it decide what valid next tokens it can generate.
28.
by ako 2mo ago
They don't need to, they can use tools to validate their assumptions.
29.
by ako 2mo ago
I think the hard part is not the syntax itself but the shift in thinking: instead of procedural state manipulation expressing the desired end result in declarative set based relation algebra. I see developers struggling with breaking down c
30.
by ako 2mo ago
Time to start using plsql, ADA with first class support for embedded SQL.

More ›