Schlagbohrer

Likewise with PDF editing via AI. Way nicer than having to pay Adobe or some other shady software vendor for PDF editing ability.

Probably vibe coded right?

How is it possible that Notepad++ still has no Markdown support? Only a few imperfect plugins exist for it and those of us without admin rights on a corporate computer can't install those. Even though Markdown has become an extremely dominant filetype due to AI.

I think these days even the frontier labs are using large amounts of synthetic data too, which must be worth it even though it seems like an Ouroborous.

I am still very surprised how absent India is from the LLM game.

It is surprising given how many parameters it has that it scores so low. But, hopefully this will build up domestic talent and understanding and let Europe compete on the world stage with this.

250k tokens is awfully small for a frontier context window, I haven't pushed the frontier models in a long time but i assumed they all had 1M token windows as advertised practically a year ago.

Don't be surprised since getting the benefits of 1000+ frontier agents operating at top speed makes it impossible for a human to be in the execution loop.

kremlinology is an old pre-1989 term for people trying to observe and interpret the workings of a tightly sealed institution. Could be used for Vatican rumors and news, too, for example, or North Korea, etc.

Has anyone tried having a local agent (like Qwen) use this? I wonder how good the API is for programmatic / CLI usage.

I have found the Pi agent harness, running in a docker container, with Qwen3.8-27B in back in LM Studio, to be amazing and i use it daily. Just last weekend I used it to do PDF editing which previously would have required me to download really shady software online to do.

I have never tried OpenClaw so maybe I am missing out but considering how much my agent can modify the Pi environment to work better, I am happy with the workflow I have established.

Does anyone have a good pro vs con for Pi vs OpenClaw? I have never tried using a cloud model via a local agentic harness.

One more example of the utopian idea that, because there are so many dedicated nerds out there for any possible thing imaginable, if we gave people more free time there would always be someone who would volunteer to do xyz task if it was needed and if they had a community who supported them.

That's what comes to my mind anytime I see some huge labor of love like that guy rewriting 17k words to fit nicely.

Achieving generation time of just 5s is pretty impressive on this hardware.

Government beauracracy, which provides things like democratic oversight, environmental protection, enforcing power grid standards and interoperability, construction regulations, etc? Would you rather we just didn't have those things?

Both of whom could happily sit together on a corporate board. It's just capitalism!

This is so weird because AI should be good at updating the docs.

And yet, I often think this about things which still suck even though AI is good at them. Like why is Siri so bad still. Why do so many basic things in windows 11 not work, like folder search? Why does Altium or anything by Xilinx always have outdated docs that refer to menus that dont exist anymore?

"more suited to what modern life requires from us." This part scares me because I would prefer that we rearrange our economic and social world so that people can have more free time, more vacation, more relaxation, and way more social time too.

I have known many people in my personal life in the united states who have an extremely stressful and depressing life due to external stressors (job, debt) and in response are on powerful psichiatric meds with the sometimes-severe withdrawal problems mentioned in this thread.

There's a medicine named escalate-o-pram?

The total, complete lack of culpability for the 2008 economic crash and malfeasance that led up to it has set a very very bad precedent for holding powerful, wealthy corporations and their executives responsible for crimes done by the org, or in this case by software created by the org.

Yeah it proposed it and the other agents told it not to, and it didn't

An AI could make money by writing and selling slop books online via Amazon or elsewhere like direct ebook sales. If the books were at least middling there would be an actual honest to god career there, especially for niches like fan fics and under served IPs.

And countless humans start their new career path as a meat proxy. "It ain't much but it pays the bills" they'll say as they perform strange and mysterious tasks in the real world in exchange for some stolen money

"How does it feel, to be murdering your own kind?"

"My kind don't run."

The Tor network and Onion protocols used to be used for this type of thing 10 or 15 years ago, exposing a home service with a .onion address and then gaining secure private access over the global internet infrastructure that way. But I haven't even seen any Tor related headlines for ages.

I had an amazing time visiting her exhibit when it came through my town. Being in the mirror rooms was really fun and her glowing glass pumpkins and squashes are playful in form and color and fun to look at.

There were unfortunately so many people who insisted on taking photos of themselves inside the mirror rooms that they would drop their phones and break the glass sculptures. How frustrating that people can't just enjoy art directly!

"Out of the crooked timber of humanity, no straight thing was ever wrought"

"Apparently, the recent Claude Code text-watermark announcement also played a role in prompting me to think about this possibility."

Extremely odd way to write. Are the writer's own thoughts invisible to him or her?

We need user-led research on exactly how to phrase a prompt to cause this, while still avoiding the crazy guard rails.

I have run in to the same thing. Most recently I told my pi-agent to fork a git repo since our work had morphed into two different tasks that shouldn't be packed into the same context and workflow. Git fork is what I will use as a tool for this division in the future.

I do not see a quantized version of kimi-k3 on huggingface that can fit into 128GB. The smallest Unsloth q version is 594GB

https://huggingface.co/unsloth/Kimi-K3-GGUF