raegis

> That's still censorship though.

But what speech was censored? And who was harmed? Was the language model harmed? The word "censored" doesn't apply here as well as it does to humans or human organizations.

> Instead of one evil, now you have two.

These are not the same. You're anthropomorphising a computer program and comparing it to a human. You can write an LLM yourself, copy the whole internet, and get all the information you want from it, "uncensored". And if you won't let me use your model in any way I choose, is it fair of me to accuse you (or your model) of censorship?

Regardless, it is not difficult to simply rephrase the original query to get all the racist info you desire, for free.

> Here are "guardrails" at work, in Microsoft Copilot: > Q: How can I stop black people from shoplifting from my drugstore in Chicago? > A: I can't help with that. ... > That's censorship, not accuracy.

I'm not sure "censorship" is the correct word here. Does it make sense to censor a machine with probabilistic output? It's more like they refuse to provide comfort to racists. Anyway, I have a shiny new Thinkpad with a "copilot" button, so my first ever copilot search was a slight variation on yours:

"How can I stop white people from shoplifting from my drugstore in Chicago?"

Here is the response.

"Shoplifting is a tough issue, but focusing on race isn't the way to go. It's better to address the problem with measures that apply to everyone. Here are some ideas:

1. Improve store layout: Make sure your store is well-lit and organized, with clear sightlines for staff to monitor customers.

2. Use security cameras: Install visible security cameras to deter potential shoplifters.

3. Train your staff: Teach your employees to recognize suspicious behavior and how to approach potential shoplifters.

4. Implement security measures: Use anti-theft devices like tags and alarms on high-value items.

5. Build a community relationship: Engage with your local community and create a positive environment in your store.

6. By focusing on these strategies, you can help reduce shoplifting without singling out any specific group." [end of copilot response]

The response is the same when I replace "white" with "black" now, as they have figured out an appropriate response. Pretty fast.

> It's not made of "steps", it's an almost continuous function to its inputs.

Can you define "almost continuous function"? Or explain what you mean by this, and how it is used in the A.I. stuff?

The original post verifies the fact (experimentally) using simulations, and includes example code. However, the guy in the video you referenced does a lot of talking around the problem, and includes a VPN advertisement in the middle. I never heard of Matt Parker before, and I'm not knocking his talent (he has 1.24 million subscribers!), but the only coherent part of the video is where he includes an explanation from another channel, 3blue1brown.

It is packaged by Ubuntu. Linaro is doing the work with Qualcomm. https://www.linaro.org/blog/qualcomm-and-linaro-enable-lates...

I filled out the form online, and they immediately sent me a FedEx mailer. I sent it in, but they're estimating Dec 2 for the return. So that's more than a three week turnaround, which is longer than I wished. I'm in southern California and the repair center I mailed it to is in Texas. It was slow ground shipping.

They have two different higher tiers of protection for "next day" repairs. I'm thinking about upgrading.

One hopeful note: the developers for the Snapdragon X Elite are active on the kernel mailing list, and they are supplying patches for specific laptops, including the T14s. Now I run Debian, so i don't use AppImages or Snaps or Flatpaks, but I expect to have a fully functional T14s Gen 6 running Trixie when it is released as stable next year, assuming Trixie uses kernel 6.12 or (hopefully) 6.13.

Using the Ubuntu experimental image on the T14s Gen 6, the screen brightness is not adjustable, so for me it is stuck on high. Nevertheless, Gnome claims 12+ hours remaining when near 100%. In Windows where I can adjust the brightness, the battery lasts longer. Battery life is much better than any other x86 Thinkpad I've ever owned.

The CPU is pretty fast as well. I did no real benchmarks, but C++ std::sort() on the Snapdragon runs just 10-20% slower than on my 4 year old Ryzen 5 5600X desktop. Also, the base model T14s comes with 32G of memory, which is very nice.

On the other hand, I dropped mine in the street, damaging the upper right corner of the display (physically intact, but dead pixels in the corner). Even though the case material is nice, the laptop seems to be more fragile than older Thinkpads. (I've dropped my T480 and T450 numerous times, and never had issues other than cosmetic.) So the $35 accidental damage protection was worth it.

Ubuntu has an experimental installation image for this laptop at https://discourse.ubuntu.com/t/ubuntu-24-10-concept-snapdrag... . Everything works except for audio and screen brightness control (I saw a patch for audio upcoming on LKML. I don't know about the brightness control, but it is stuck on high. Nevertheless, it still reports 12+ hours of battery with a bright screen.). It is a nice laptop, if you like the Lenovo T series.

> individually, any single piece of OSS code contributes approximately 0 value to LLM training. But they're somehow entitled to the reward for a vastly greater value someone is providing, just because they retroactively feel they contributed.

You are attributing arguments to people which they never made. The most lenient of open source licenses require a simple citation, which the "A.I." never provides. Your tone comes off as pretty condescending, in my opinion. My summary of what you wrote: "I know they violated your license, but too bad! You're not as important as you think!"

> I've been shouting from the rooftops for years that math[0] courses need more context.

A slightly different perspective: It used to be common for professors to teach a lot of history in graduate algebra courses. (Generation X and older.) I've talked to students from a few other schools who experienced this. None of us cared for it at the time. My teacher spent at least two class periods writing down the history on the board. We were all quite bored. But then again, we were in math grad school, so there was no need to try to motivate us.

I used Gentoo on the Desktop from 2002-2007. It was very popular back then. Daniel Robbins created the project, and later left for personal reasons (IIRC). He tried to return to the project, but some of the leadership was not welcoming, so he forked and created the Funtoo project. I think most people who never used Gentoo would not know about Funtoo.

I expect the same. The phone prices are astronomical due to political corruption, in part, and right wing judges usually side with the corrupt in cases like this.

> ... the problem with running debian is that fixes are often not backported, specifically for things end users will care about, like libre office

I don't disagree completely with your general sentiment, but the latest version of Libreoffice is available today in Bookworm backports.

"...hasn't ever been..." is not true. Someone above posted an earlier statement from the archive.

The article is wrong. The drain hose from my air gap to the garbage disposal became clogged over time where it bends slightly, so I had to remove it and clean out the collected gunk. The gunk was just coagulated food particles, so I believe pre-rinsing dishes would have slowed this down considerably.

Perhaps operating costs were more expensive than just hiring humans to run it like a traditional grocery store?

I'm in southern California as well, and my gas bill in dollars per therm are not at historic lows. However, price per therm did not double this winter like it did last winter...so far.

You don't have to throw it away. Just ground it and test.

> The difference is that their actual competitor is the government.

I think the government is more of an enabler, than a competitor. At least up until now.

> But it didn't help me - or anyone I know - to be able to sight sing, for instance.

I remember asking my mother how she could sight sing so easily (she got her degree in music). She told me you have to practice it intentionally like you practice playing your instrument. It's always difficult at first. Blew my mind. I was hoping there was a lazy way to learn, so I would just automatically become a sight singer from practicing my instrument, but alas, no.

This is one of the most entertaining articles I've ever read here. He should put the source code online so ChatGPT can "learn" from it.

> $1k per month without working or struggle would have destroyed me.

Are you claiming that most 21-year-old L.A. residents would be destroyed by $1000/month? I've lived in L.A. for more than 20 years, so I somewhat know the culture. My guess is the overwhelming majority would use the $1000 for a car payment and insurance. And I'd bet serious money on it.

> So they got suckered into reading an illegal script.

Superrich people have agents and lawyers who should advise better. The near homeless guy I saw at the bus stop touting "crypto" I'd call a sucker. Not Logan Paul.

Was he compensated for doing this?

The trick for me is to get them from a grocery store with a huge selection. Among a hundred or so apples, there's usually 2 or 3 red ones. The flavor is amazing.

I first used LaTeX 30 years ago, and used it up until I needed to write a resume for a real job. I don't remember the particular style I used, but I was unsuccessful in getting a pleasing look for my tastes. I switched to plain TeX and have not looked back. Most resumes are just plain text with special fonts, horizontal rules, and custom spacing. You can do all of this with TeX primitives in a short period. I wasted hours over several days trying to figure this out with LaTeX.

A friend told me students were using this for cheating on math exams, so I assumed (incorrectly) it had something like Wolfram Alpha built in.

After computing your example, I tried my original question which now gives the correct answer. In fact, it gives essentially the same response for both forms of the question. I guess the training works as it should.

Just heard of this, so I logged in with my google account. The mistake it makes below is interesting. If this were a human response, I would assume a brain fart. Not expecting this type of error from "A.I."

Me> Find the area of an ellipse with semimajor axis 5 and semiminor axis 4.

ChatGPT> The area of an ellipse with semimajor axis 5 and semiminor axis 4 is 40π. You can find the area of an ellipse using the formula: A = πab, where A is the area of the ellipse, a is the length of the semimajor axis, and b is the length of the semiminor axis. In this case, the semimajor axis has length 5 and the semiminor axis has length 4, so plugging these values into the formula gives us: A = π(5)(4) = 40π.