iiinova — where your digital self keeps evolving

Bram Zijlstra @bramzijlstra.com · 02/04/2025

Every now and then I read posts about 90% of coding will be done by AI this year and I can't lie it makes me a bit nervous. Then a study likes this drops and I can relax a bit

140

Reposted by Bram Zijlstra

Sung Kim @sungkim.bsky.social · 31/03/2025

You’ve probably heard about how AI/LLMs can solve Math Olympiad problems ( deepmind.google/discover/blo... ). So naturally, some people put it to the test — hours after the 2025 US Math Olympiad problems were released. The result: They all sucked!

917350

Bram Zijlstra @bramzijlstra.com · 29/03/2025

There are plenty of cases where survivorship bias doesn’t apply. We just don’t remember them.

120

Bram Zijlstra @bramzijlstra.com · 26/03/2025

Errors with a 200 status code is like gift wrapping a turd

000

Bram Zijlstra @bramzijlstra.com · 25/03/2025

An infinite amount of monkeys with an LLM can write the entire the complete works of Shakespeare

110

Bram Zijlstra @bramzijlstra.com · 20/03/2025

My doctor told me he's into vibe surgeries lately

060

Bram Zijlstra @bramzijlstra.com · 18/03/2025

I think the reason why LLMs are overconfident is because we keep saying telling "You are an expert in" literally anything

130

Bram Zijlstra @bramzijlstra.com · 03/03/2025

Electron apps are the Monkey's Paw to someone wishing cheaper RAM

000

Bram Zijlstra @bramzijlstra.com · 28/02/2025

Anyone know best practices / tips for improving LLM quality on classifying long texts? For shorter input few-shot learning and finding the best examples work well, but this not very practical with longer texts. #databs

020

Bram Zijlstra @bramzijlstra.com · 24/02/2025

We'll have AI alignment before MS Word alignment

040

Bram Zijlstra @bramzijlstra.com · 18/02/2025

An agent is an LLM that uses tools. A tool is someone who keeps saying '2025 is the year of the agents'

010

Bram Zijlstra @bramzijlstra.com · 07/02/2025

Today was a day where all emails found me well

020

Bram Zijlstra @bramzijlstra.com · 07/02/2025

Interesting benchmark from Adyen on multi-step reasoning. New benchmarks are great in establishing a baseline for historic models, all new models should be treated with suspicion. #databs huggingface.co/blog/dabstep

021

Bram Zijlstra @bramzijlstra.com · 07/02/2025

Really want to like cursor but I am completely Pycharm brained it seems. Anyone else have the same ? Which copilot did you go for?

010

Bram Zijlstra @bramzijlstra.com · 06/02/2025

I'm always surprised that ChatGPT suggests old school NLP / string manipulation tricks instead of suggesting to call OpenAI. Seems like upselling 101 to me.

110

Bram Zijlstra @bramzijlstra.com · 04/02/2025

“Our company has moat” The moat:

020

Bram Zijlstra @bramzijlstra.com · 03/02/2025

DeepSeek is not refusing the Tiananmen Square question, though it gets the answer wrong.

Screenshot of DeepSeek chat.
User: How many r's are in the words 'Tiananmen Square'?
DeepSeek: The phrase "Tiananmen Square" contains 2 r's.
If you have any other questions, feel free to ask!

111

Reposted by Bram Zijlstra

Mike Masnick @masnick.com · 29/01/2025

So, look. I'm sure I'm in the minority here on Bluesky in believing that training AI systems isn't copyright infringement. But, also. Dude. There's no way OpenAI can make this argument without looking very, very silly.

3847447

Bram Zijlstra @bramzijlstra.com · 27/01/2025

Will copilots (eventually) negatively impact open source library development? Major updates will break any copilot, so it disincentives doing so.

220

Reposted by Bram Zijlstra

Gergely Orosz @gergely.pragmaticengineer.com · 27/01/2025

Baffling to see the disclosure around DeepSeek Last I checked the tech industry, we celebrate small teams pushing the industry forward, coming up w novel ways to build software more efficiently. And sharing it. Yet now there's a class of folks who think this is some bad thing?

4361957

Bram Zijlstra @bramzijlstra.com · 27/01/2025

If I would be a dictator for a day I would ban the word 'syncing' outside the computer domain.

010

Reposted by Bram Zijlstra

Simon Willison @simonwillison.net · 20/01/2025

DeepSeek released a whole family of inference-scaling / "reasoning" models today, including distilled variants based on Llama and Qwen Here are my notes on the new models, plus how I ran DeepSeek-R1-Distill-Llama-8B on my Mac using Ollama and LLM simonwillison.net/2025/Jan/20/...

simonwillison.net

DeepSeek-R1 and exploring DeepSeek-R1-Distill-Llama-8B

DeepSeek are the Chinese AI lab who dropped the best currently available open weights LLM on Christmas day, DeepSeek v3. That model was trained in part using their unreleased R1 …

1520331

Reposted by Bram Zijlstra

Hardik @hardikvala.com · 21/01/2025

Lol, if only we could give R1 a hug.

021

Bram Zijlstra @bramzijlstra.com · 21/01/2025

I am working with Deepseek R1 today and used it for a simple classifier with few-shot learning. Looking at the thinking tokens I think I know now how anxiety and extreme perfectionism looks like

240

Bram Zijlstra @bramzijlstra.com · 20/01/2025

When you replace cheese on your pizza with feces you can't claim to save money on ingredients, you just make everything taste like shit.

010

Bram Zijlstra @bramzijlstra.com · 20/01/2025

Anyone else feel really proud when o1 needs to think a long time for your prompt?

020

Bram Zijlstra @bramzijlstra.com · 18/01/2025

Google should create a "no captcha" game where you have to pretend to be a robot and get points for tricking the algorithm

010

Bram Zijlstra @bramzijlstra.com · 15/01/2025

If your AI boyfriend is sharing weights with other women, is that considered cheating?

020

Bram Zijlstra @bramzijlstra.com · 15/01/2025

Do other things have a 'Bitter Lesson' like Machine Learning? Restaurants: It's always the salt and butter that makes it better Stress: It's finishing the task you dread most that will reduce your anxiety

220

Bram Zijlstra @bramzijlstra.com · 15/01/2025

“If you could walk a mile in my shoes” is the Docker equivalent of life experiences

110

Bram Zijlstra @bramzijlstra.com · 14/01/2025

Quote from December 2024. Not even a month later someone trains an open-source o1 model for under 500 bucks. Funny how fast LLM takes are outdated.

Screenshot of a quote:
"Basically, a frontier model like OpenAI’s O1 is like a Ferrari SF-23. It’s an obvious triumph of engineering, designed to win races, and that’s why we talk about it. But it takes a special pit crew just to change the tires and you can’t buy one for yourself. In contrast, a BERT model is like a Honda Civic. It’s also an engineering triumph, but more subtly, since it is engineered to be affordable, fuel-efficient, reliable, and extremely useful. And that’s why they’re absolutely everywhere."

120

Bram Zijlstra @bramzijlstra.com · 13/01/2025

My running coach says I’m not a real runner until I fix my form. I think he's gait keeping

041

Reposted by Bram Zijlstra

Our World in Data @ourworldindata.org · 10/12/2024

The world is awful. The world is much better. The world can be much better. All three statements are true at the same time. Understanding this is key to solving big global problems. We believe data & research can help us understand both the problems we face & the progress that’s possible. 🧵

A chart in three parts showing data on child mortality to make the points that "The world is awful. The world is much better. The world can be much better. All three statements are true at the same time."

111241416

Bram Zijlstra @bramzijlstra.com · 06/01/2025

When the chips in small size Make the compute go twice, that's a Moore #databs

010

Reposted by Bram Zijlstra

mr. TIM @timkellogg.me · 06/01/2025

my hot take: data IS NOT better than pure vibes because it has a way of causing people’s brains to shut off. i think a hallmark trait of the currently growing generation alpha will be their skepticism, questioning everything they see this is the way. data is not enough

341

Bram Zijlstra @bramzijlstra.com · 06/01/2025

I've been tracking my stress levels on Garmin and a stupid tip is just to stress less. It works 🤷‍♂️

000

Bram Zijlstra @bramzijlstra.com · 06/01/2025

Tired: Saying "It's so over" after every AI breakthrough Wired: Saying "It's so over" in normal conversations every time you hear something new

010

Bram Zijlstra @bramzijlstra.com · 04/01/2025

I thought buying an M2 would set me for years but seeing people using GPT-4 like models on M4's make me feel like I'm programming on a wet cardboard.

030

Bram Zijlstra @bramzijlstra.com · 03/01/2025

Normalise interpersonal consensus forming on what is socially acceptable or not

010

Reposted by Bram Zijlstra

🏴‍☠️ @jik.wtf · 03/01/2025

trending topics is a great way to break out of my echo chamber man things seem bleak outside my echo chamber

1273

Bram Zijlstra @bramzijlstra.com · 03/01/2025

I once took a course in Heideggers "Seind Un Zeit" and I kept changing my mind between "this is genius and revolutionary" and "this is completely incomprehensible", often multiple times per page and without a definitive conclusion

120

Reposted by Bram Zijlstra

mr. TIM @timkellogg.me · 03/01/2025

you can spot a hobby project on the internet because it uses 100% IPv6

041

Bram Zijlstra @bramzijlstra.com · 03/01/2025

The correlation between how much I like my post and how well it is doing is not a positive number

030

Bram Zijlstra @bramzijlstra.com · 03/01/2025

Affirmations are prompt instructions for humans

030

Bram Zijlstra @bramzijlstra.com · 03/01/2025

2005: You can't trust anything that's on Wikipedia 2025: You can't trust anything besides Wikipedia

3323155

Bram Zijlstra @bramzijlstra.com · 03/01/2025

Having a frontend developer in your team is like having a very attractive friend in terms of attention

050

Bram Zijlstra @bramzijlstra.com · 01/01/2025

Reddit used to be a great source for regional information (i.e. VAT regulation in the Netherlands), but now they sneakily auto translate pages that you find on Google. Happens quite frequently that a few comments in you notice that it doesn't apply to you. Super annoying

141