Haven’t tried Gemini; may work. But, in my experience with other LLMs, even if text doesn’t exceed the token limit, LLMs start making more mistakes and sometimes behave strangely more often as the size of context grows.

I usually just use VS Code to do full-text searches, and write down notes in a note taking app. That, and browse the documentation.

Nah, LLMs have severe context window limitations. It starts to get wackier after ~1000 LOC.

Python is quite slow, so will use more CPU cycles than many other languages. If you’re doing data-heavy stuff, it’ll probably also use more RAM than, say C, where you can control types and memory layout of structs.

That being said, for services, I typically use FastAPI, because it’s just so quick to develop stuff in Python. I don’t do heavy stuff in Python; that’s done by packages that wrap binaries complied from C, C++, Fortran, or CUDA. If I need tight-loops, I either entirely switch to a different language (Rust, lately), or I write a library and interact with it with ctypes.

Production AI is highly tuned by training data selection and human feedback. Every model has its own style that many people helped tune. In the open model world there are thousands of different models targeting various styles. Waifu Diffusion and GPT-4chan, for example.

I think you have your janitor example backwards. Spending my time revolutionizing energy productions sounds much more enjoyable than sweeping floors. Same with designing an effective floor sweeping robot.

AI are people, my friend. /s

But, really, I think people should be able to run algorithms on whatever data they want. It’s whether the output is sufficiently different or “transformative” that matters (and other laws like using people’s likeness). Otherwise, I think the laws will get complex and nonsensical once you start adding special cases for “AI.” And I’d bet if new laws are written, they’d be written by lobbiests to further erode the threat of competition (from free software, for instance).

The search engine LLMs suck. I’m guessing they use very small models to save compute. ChatGPT 4o and Claude 3.5 are much better.

C# is actually pretty nice. Ecosystem, not so much, but D doesn’t really have one anyways :)

Yeah, the image bytes are random because they’re already compressed (unless they’re bitmaps, which is not likely).

Donation, patronage, gift economy, mutual aid, or whatever you want to call it is fine by me. People can pirate a lot of proprietary software as well, yet people still pay.

Yet, people still pay for it.

The problem is that HP writes drivers and software for those things for Windows, but not for Linux, so Linux depends on random people to write software for those things for free (which often involves complex reverse-engineering). With Linux you need to make sure you use widely-used hardware that someone has already written support for (this is mostly applicable to laptops and peripherals, which often use custom non-standard hardware). There may be a way to fix your problems, but you’ll have to search forums or issue trackers for the solutions, and they’re probably pretty involved to get working correctly. The router crashing thing is probably just a coincidence though, or the laptop is using a feature that’s broken on your router.

If you’re talking about naive bayes filtering, it most definitely is an ML model. Modern spam filters use more complex ML models (or at least I know Yahoo Mail used to ~15 years ago, because I saw a lecture where John Langford talked a little bit about it). Statistical ML is an “AI” field. Stuff like anomaly detection are also usually ML models.

OSMC’s Vero V looks interesting. Pi 4 with OSMC or Librelec could work. I’m probably going to do something like this pretty soon. I just set up an *arr stack last week, and just using my smart TV with the jellyfin app installed ATM.

My PC running the Jellyfin server can’t transcode some videos though; probably going to put an Arc a310 in it.

In the Texas counties I’m most familiar with, if you’re arrested and they don’t have a good case, they just keep resetting court dates for years instead of going ahead with the process. If you can’t afford a bond, you’ll be in jail that whole time (which pressures people to take plea deals), if you can secure a bond, you’re out, but with limited rights and a whole lot of hassles to deal with.

I’ve used them as a proxy for a web app at the last place I worked. Was just hoping they’d block unwanted/malicious traffic (not sure if it was needed, and it wasn’t my choice). I, personally, didn’t have any problems with their service.

Now, if you take a step back, and look at the big picture, they are so big and ubiquitous that they are a threat to the WWW itself. They are probably one of the most valuable targets for malicious actors and nation states. Even if Cloudflare is able to defend against infiltration and attacks in perpetuity, they have much of the net locked-in, and will enshittify to keep profits increasing in a market they’ve almost completely saturated.

Also, CAPTCHAs are annoying.

I think similar, and arguably more fine-grained, things can be done with Typescript, traditional OOP (interfaces, and maybe the Facade pattern), and perhaps dependency injection.

I thought the tuning procedures, such as RLHF, kind of messes up the probabilities, so you can’t really tell how confident the model is in the output (and I’m not sure how accurate these probabilities were in the first place)?

Also, it seems, at a certain point, the more context the models are given, the less accurate the output. A few times, I asked ChatGPT something, and it used its browsing functionality to look it up, and it was still wrong even though the sources were correct. But, when I disabled “browsing” so it would just use its internal model, it was correct.

It doesn’t seem there are too many expert services tied to ChatGPT (I’m just using this as an example, because that’s the one I use). There’s obviously some kind of guardrail system for “safety,” there’s a search/browsing system (it shows you when it uses this), and there’s a python interpreter. Of course, OpenAI is now very closed, so they may be hiding that it’s using expert services (beyond the “experts” in the MOE model their speculated to be using).

I find Kagi results a little bit better than Google’s (for most things). I like that certain categories of results are put in their own sections (listicles, forums) so they’re easy to ignore if you want. I like that I can prioritize, deprioritize, block, or pin results from certain domains. I like that I can quickly switch “lenses” to one of the predefined or custom lenses.