HNHacker News
TopNewBestAskShowJobs

Glemkloksdjf

36 karma · joined November 25, 2025

submissionscomments
Glemkloksdjf··on I don't care how well your "AI" works
If you are a software engineere, you can leverage AI a lot better to write code than anyone else.

The complexity of good code, is still complicated.

which means 1. if software development is really solved, everyone else also gets a huge problem (ceo, cto, accountants, designers, etc. etc.) so we are in the back of the ai doomsday line.

And 2. it allows YOU to leverage AI a lot better which can enable you to create your own product.

In my startup, we leverage AI and we are not worried that another company just does the same thing because even if they do, we know how to write good code and architecture and we are also using AI. So we will always be ahead.

Glemkloksdjf··on I don't care how well your "AI" works
Every good 'techie' around me has it good.
Glemkloksdjf··on Copyparty, the FOSS file server [video]
I tried it a few month ago. I wanted to move 2tb of different type of data from one pc to a server.

The upload of these files were horrendes, it was slow on computing something, it was slow in transfering and the ui sucked.

I enabled sshd and moved everything over with scp.

I loved the video, i hated my one time experience with it :|

Glemkloksdjf··on I don't care how well your "AI" works
Its ignorant. Thats what it is.

The big tech will build out compute in a never seen speed and we will reach 2e29 Flops faster than ever.

Big tech is competing with each other and they are the ones with the real money in our capitalistic world but even if they would find some slow down between each others, countries are also now competing.

In the next 4 years and the massive build out of compute, we will see a lot clearer how the progress will go.

And either we hit obvous limitations or not.

If we will not see an obvious limitation, fionas opinion will have 0 relevance.

The best chance for everyone is to keep a very very close eye on AI to either make the right decisions (not buying that house with a line of credit; creating your own product a lot faster thanks to ai, ...) or be aware what is coming.

Thanks for the fish and enjoy the ride.

Glemkloksdjf··on Orion 1.0
Its not a new browser, its a new webengine wrapper?
Glemkloksdjf··on Meta Segment Anything Model 3
You can get labels out of the classifier and bounding box models.

They are super fast.

Its just an alternative i'm mentioning. I would assume a person knowing a little bit of that domain.

Otherwise the first option would be CLIP i assume. llm-vl is just super slow and compute intensive.

Glemkloksdjf··on Meta Segment Anything Model 3
I suggested YOLO and non llm-vl as a lot faster alternative.

Of course CLIP would be otherwise the other option than a big llm-vl one.

Glemkloksdjf··on Meta Segment Anything Model 3
Any current one. they are easy to use and you can just benchmark them yourself.

I'm using small and medum.

Also the code for using it is very short and easy to use. You can also use ChatGPT to generate small exepriments to see what fits your case better

Glemkloksdjf··on Meta Segment Anything Model 3
I would suggest YOLO. Depending on your domain, you might also finetune these models. Its relativly easy as they are not big LLMs but either image classification or bounding boxes.

I would recommend bounding boxes.

Glemkloksdjf··on Three Years from GPT-3 to Gemini 3
So you don't even try LLMs regularly?

And whats with everything else regarding ML progress like image generation, 3d world generation etc.?

I vibe coded plenty of small things i haven't ever had the time for them. You don't have anything which you wanted to do and can fit in a single page html application? It can even use local storage etc.

Glemkloksdjf··on The Bitter Lesson of LLM Extensions
No, skills are a set of manifested and tested 'skills' which reduce the 'mental load' of the LLM and reduces the context the LLM needs to do things reproducable.

Similiar to what humans do.

← PreviousPage 3 of 3