Karsten Konrad 🇪🇺
@zabong69.bsky.social
Armchair Data Scientist and nerd. PhD in automated theorem proving just before AI became a subfield of Linear Algebra. Machine Learning, statistics, data visualisation, biology, recreational maths, computation, old pocket calculators, and video games
Reposted by Karsten Konrad 🇪🇺
Looking over my left shoulder, I see Mom staring at me, knuckles white on the handle of her cart. I turn my head to find Nana behind the avocados, glaring. My heart pounding, I raise a hand towards the shelf. A bead of sweat falls from my forehead, hitting the floor with a splat.
January 2, 2026 at 2:25 PM
Looking over my left shoulder, I see Mom staring at me, knuckles white on the handle of her cart. I turn my head to find Nana behind the avocados, glaring. My heart pounding, I raise a hand towards the shelf. A bead of sweat falls from my forehead, hitting the floor with a splat.
Reposted by Karsten Konrad 🇪🇺
Ikigai - that interesting self-development concept & prism from Japan - and it's (more realistic?) opposite (by @adamgrant)
January 1, 2026 at 7:55 PM
Ikigai - that interesting self-development concept & prism from Japan - and it's (more realistic?) opposite (by @adamgrant)
Reposted by Karsten Konrad 🇪🇺
Wir schauen zu - bei Photovoltaik, Batterien, e-Autos, Schnellzügen, intelligentem Feuerwerk, KI insgesamt, uvm - wie China an Europa technisch vorbeizieht. Wie lange noch? Resilienz, Standortsicherung & Autonomie heißt auch, technisch mitzuhalten statt an Altem festzuhalten. 2026 muss wachrütteln.
January 1, 2026 at 11:54 AM
Wir schauen zu - bei Photovoltaik, Batterien, e-Autos, Schnellzügen, intelligentem Feuerwerk, KI insgesamt, uvm - wie China an Europa technisch vorbeizieht. Wie lange noch? Resilienz, Standortsicherung & Autonomie heißt auch, technisch mitzuhalten statt an Altem festzuhalten. 2026 muss wachrütteln.
Reposted by Karsten Konrad 🇪🇺
11 year old just told me they are learning about quantitative and qualitative data and independent and dependent variables in science class and I emitted a high pitched squeal of joy
December 29, 2025 at 4:36 PM
11 year old just told me they are learning about quantitative and qualitative data and independent and dependent variables in science class and I emitted a high pitched squeal of joy
Reposted by Karsten Konrad 🇪🇺
As a data engineer who learned to code long before AI and uses Claude daily to assist me code and learn new tools, I think AI was designed for autists like me.
I honestly don't know many other good use cases for AI beyond coding, survey-level research, and interrogating specific documents.
I honestly don't know many other good use cases for AI beyond coding, survey-level research, and interrogating specific documents.
December 29, 2025 at 7:15 AM
As a data engineer who learned to code long before AI and uses Claude daily to assist me code and learn new tools, I think AI was designed for autists like me.
I honestly don't know many other good use cases for AI beyond coding, survey-level research, and interrogating specific documents.
I honestly don't know many other good use cases for AI beyond coding, survey-level research, and interrogating specific documents.
Reposted by Karsten Konrad 🇪🇺
So, Kirsten hat sich nun wieder auf den Weg gemacht. Ich lüfte erstmal durch. Wissen Se, ihre Tees und Kräuter riechen doch recht streng. Am schlimmsten ist dieses Marie-Johanna!
December 28, 2025 at 11:49 AM
So, Kirsten hat sich nun wieder auf den Weg gemacht. Ich lüfte erstmal durch. Wissen Se, ihre Tees und Kräuter riechen doch recht streng. Am schlimmsten ist dieses Marie-Johanna!
Reposted by Karsten Konrad 🇪🇺
The state of programming in 2025. Anyways, off to write a CLAUDE•md…
December 26, 2025 at 8:32 PM
The state of programming in 2025. Anyways, off to write a CLAUDE•md…
Reposted by Karsten Konrad 🇪🇺
Using AI coding for data analysis without personal programming skill fills me with dread.
Small errors in the code poisons results in ways that may not be visibly obvious.
LLMs are great when people verify outputs; the path to hell is when they don't.
Small errors in the code poisons results in ways that may not be visibly obvious.
LLMs are great when people verify outputs; the path to hell is when they don't.
December 26, 2025 at 5:07 PM
Using AI coding for data analysis without personal programming skill fills me with dread.
Small errors in the code poisons results in ways that may not be visibly obvious.
LLMs are great when people verify outputs; the path to hell is when they don't.
Small errors in the code poisons results in ways that may not be visibly obvious.
LLMs are great when people verify outputs; the path to hell is when they don't.
Reposted by Karsten Konrad 🇪🇺
I am an AI researcher. I love AI. It’s so cool.
I do not love companies that force AI upon me so that they can post impressive sounding data center stats.
All I want is an easy way to opt out. I’ll let you know when I want to use AI. I promise.
I do not love companies that force AI upon me so that they can post impressive sounding data center stats.
All I want is an easy way to opt out. I’ll let you know when I want to use AI. I promise.
December 16, 2025 at 5:00 PM
I am an AI researcher. I love AI. It’s so cool.
I do not love companies that force AI upon me so that they can post impressive sounding data center stats.
All I want is an easy way to opt out. I’ll let you know when I want to use AI. I promise.
I do not love companies that force AI upon me so that they can post impressive sounding data center stats.
All I want is an easy way to opt out. I’ll let you know when I want to use AI. I promise.
In the spirit of Christmas, let‘s just say if you bring up arguments against AI-by-learning that make use of intractability arguments, make sure that your argument doesn’t also apply to well-known use cases that already work very well, like image classification.
It's deeply unserious to be pushing rhetoric _like this_ over a publication that has unaddressed (for a year and counting now!) issues that were formally brought up in a published critique! The entire backing proof falls apart!
It's downright unethical. Do better.
It's downright unethical. Do better.
December 26, 2025 at 9:53 AM
In the spirit of Christmas, let‘s just say if you bring up arguments against AI-by-learning that make use of intractability arguments, make sure that your argument doesn’t also apply to well-known use cases that already work very well, like image classification.
Claude Code is happily building a data viz web app for me all day. Incredible that this didn't even exist 11 months ago.
December 24, 2025 at 1:33 PM
Claude Code is happily building a data viz web app for me all day. Incredible that this didn't even exist 11 months ago.
Reposted by Karsten Konrad 🇪🇺
NEW: last night, the US govt launched an assault on the ‘global censorship industrial complex’, aka European tech researchers/campaigners.
It’s a deeply chilling move & speaks as to why we need international solidarity more than ever.
open.substack.com/pub/broligar...
It’s a deeply chilling move & speaks as to why we need international solidarity more than ever.
open.substack.com/pub/broligar...
A message to America: we are not your enemy
Last night, the US launched an all-out assault on the "global censorship industrial complex", I respond with some deep breaths, solidarity and a vibey video (it's all I have)
open.substack.com
December 24, 2025 at 11:52 AM
NEW: last night, the US govt launched an assault on the ‘global censorship industrial complex’, aka European tech researchers/campaigners.
It’s a deeply chilling move & speaks as to why we need international solidarity more than ever.
open.substack.com/pub/broligar...
It’s a deeply chilling move & speaks as to why we need international solidarity more than ever.
open.substack.com/pub/broligar...
Reposted by Karsten Konrad 🇪🇺
Ein Dankeschön an alle, die an #Weihnachten arbeiten:
In Krankenhäusern
In Altenheimen
In Elektrizitätswerken
Bei der Polizei
Bei der Feuerwehr
In der Seelsorge
An den Tankstellen
In den Hotels und Restaurants
In den Redaktionsstuben
Beim Militär
In Rechenzentren
Uvm
In Krankenhäusern
In Altenheimen
In Elektrizitätswerken
Bei der Polizei
Bei der Feuerwehr
In der Seelsorge
An den Tankstellen
In den Hotels und Restaurants
In den Redaktionsstuben
Beim Militär
In Rechenzentren
Uvm
December 24, 2025 at 7:34 AM
Ein Dankeschön an alle, die an #Weihnachten arbeiten:
In Krankenhäusern
In Altenheimen
In Elektrizitätswerken
Bei der Polizei
Bei der Feuerwehr
In der Seelsorge
An den Tankstellen
In den Hotels und Restaurants
In den Redaktionsstuben
Beim Militär
In Rechenzentren
Uvm
In Krankenhäusern
In Altenheimen
In Elektrizitätswerken
Bei der Polizei
Bei der Feuerwehr
In der Seelsorge
An den Tankstellen
In den Hotels und Restaurants
In den Redaktionsstuben
Beim Militär
In Rechenzentren
Uvm
"We need to buy something much larger, the costs on the small thing are overrunning" will come in very handy in my future negotiations about houshold investments.
President Trump claims the warship will be "the fastest, the biggest, and by far 100 times more powerful than any battleship ever built." A month ago, the Navy scrapped plans to build a new, small warship, citing delays and cost overruns. n.pr/4pSg7zB
Trump announces plans for new Navy 'battleship' as part of a 'Golden Fleet'
President Trump claims the warship will be "the fastest, the biggest, and by far 100 times more powerful than any battleship ever built." A month ago, the Navy scrapped plans to build a new, small warship, citing delays and cost overruns.
n.pr
December 23, 2025 at 3:07 PM
"We need to buy something much larger, the costs on the small thing are overrunning" will come in very handy in my future negotiations about houshold investments.
Reposted by Karsten Konrad 🇪🇺
awesome github project that can help to get better with claude code github.com/luongnv89/c...
GitHub - luongnv89/claude-howto: Complete collection of examples for some important Claude Code features and concepts.
Complete collection of examples for some important Claude Code features and concepts. - luongnv89/claude-howto
github.com
December 23, 2025 at 11:00 AM
awesome github project that can help to get better with claude code github.com/luongnv89/c...
Pythagoras is just Big Math pushing for more RIGHT angles. Some journalist should investigate.
December 23, 2025 at 7:33 AM
Pythagoras is just Big Math pushing for more RIGHT angles. Some journalist should investigate.
Wise words.
my simple two-step system for decluttering:
1. organize items into logical piles
2. transfer piles to incinerator
1. organize items into logical piles
2. transfer piles to incinerator
December 22, 2025 at 4:15 PM
Wise words.
I am very puzzled by „LLMs are not (part of) machine learning“ by people who claim to have expertise.
December 22, 2025 at 1:29 PM
I am very puzzled by „LLMs are not (part of) machine learning“ by people who claim to have expertise.
Reposted by Karsten Konrad 🇪🇺
Ideally we should have just taken "it can interpolate over all human knowledge in natural language" and just sat with that for a few years.
It really is a gob-smacking capability, but we didn't get a moment to appreciate it before drowning it in 31 flavors of bullshit hype.
It really is a gob-smacking capability, but we didn't get a moment to appreciate it before drowning it in 31 flavors of bullshit hype.
December 21, 2025 at 5:55 PM
Ideally we should have just taken "it can interpolate over all human knowledge in natural language" and just sat with that for a few years.
It really is a gob-smacking capability, but we didn't get a moment to appreciate it before drowning it in 31 flavors of bullshit hype.
It really is a gob-smacking capability, but we didn't get a moment to appreciate it before drowning it in 31 flavors of bullshit hype.
Starting with an empty folder, today I created a command line data analysis tool that can read data and perform both standard analyses and custom ones using a local coding LLM. With Claude code, not a single problem. Code needs some refactoring, but is far from ugly.
December 19, 2025 at 2:17 PM
Starting with an empty folder, today I created a command line data analysis tool that can read data and perform both standard analyses and custom ones using a local coding LLM. With Claude code, not a single problem. Code needs some refactoring, but is far from ugly.
Reposted by Karsten Konrad 🇪🇺
Tech executives saying something stupid is really not an excuse for turning your brain off and not being able to evaluate a technology on its own terms
December 19, 2025 at 5:40 AM
Tech executives saying something stupid is really not an excuse for turning your brain off and not being able to evaluate a technology on its own terms
Reposted by Karsten Konrad 🇪🇺
"Riding in upon a teal Ford Focus came a great warrior, a suitor of the gentlefolks' granddaughter. Word had spread through the kingdom that this warrior worked with computers and perhaps even knew the true nature of the Router."
In Which I Fix My Girlfriend’s Grandparents’ WiFi and Am Hailed as a Conquering Hero
Lo, in the twilight days of the second year of the second decade of the third millennium did a great darkness descend over the wireless internet co...
buff.ly
December 18, 2025 at 2:20 AM
"Riding in upon a teal Ford Focus came a great warrior, a suitor of the gentlefolks' granddaughter. Word had spread through the kingdom that this warrior worked with computers and perhaps even knew the true nature of the Router."
Reposted by Karsten Konrad 🇪🇺
Here’s to a merry Christmas, if that’s a thing.
Tickets at: jonathanpie.com
Tickets at: jonathanpie.com
December 18, 2025 at 2:46 PM
Here’s to a merry Christmas, if that’s a thing.
Tickets at: jonathanpie.com
Tickets at: jonathanpie.com
I am running a local coding LLM and oh boy is my poor laptop fighting not to overheat. But it does work…
December 18, 2025 at 3:00 PM
I am running a local coding LLM and oh boy is my poor laptop fighting not to overheat. But it does work…
My thoughts exactly.
everyone who uses an llm to code more than me is voluntarily deskilling themselves, everyone who uses it less than me might as well be hand-wiring together vacuum tubes
December 18, 2025 at 7:18 AM
My thoughts exactly.