Promiseland

[Fetian]

I'm a few pages into the toki pona lesson plan I'm using (apparently I'm on page 5) -- enough to be able to construct simple sentences like 'I'm asleep' (mi lape) and 'you're funny' (sina musi) and 'they're eating an apple' (ona li moku e kili loje), and then I decided that I wanted to swap to focusing just on vocabulary. I've whizzed through that (I have ₇₀ words/symbols memorised, I expect to be done with the game I'm using to study it within the next couple days -- I don't think this has all words currently in active use but it'll give me the core 120-130 or so). The grammar was a little frustrating to have to keep stopping reading to go remind myself which word means what while doing the exercises, so swapping back when I'm done with the vocab will let it go a bit smoother, I think. I am enjoying myself, though!

 
Also for anyone wondering, '[sina] o moku e palisa unpa' is 'eat a dick'

[Fetian]

  Hidden text. You are not a member of the groups to which this text is available

[Fetian]

Pretty good day! Up early to go to the store with ash, where I stocked up on some things I've been needing. They were out of my usual taquitos so I had to get chicken instead, and while they had blood oranges and cara cara oranges again, when I got home I discovered the blood oranges were not particularly ripe, which is disappointing. Also they keep being very low on or entirely out of the beef baby food, which is kind of weird. But yeah

I also did, as predicted, finish off the game I was using to learn the toki pona core lexicon, so now I have a pretty good grasp of 121 words and should be able to move on back to the grammar lessons. I will probably be able to add any other words through exposure

ash and I watched Sandman: Death: The High Cost of Living, which means we are officially done with Sandman. Official final verdict: Season 2 is bad, High Cost of Living is good.

We also finished off the video series we've been watching very very slowly of a Japanese man reacting to Ludwig (youtuber/streamer) vlogging himself (and Michael Reeves, also youtuber/streamer) travelling across the entirety of Japan on a motorcycle with no map and very little Japanese between them. I'd watched Ludwig's series ages ago but the reaction series abridges it nicely and it's fun seeing the reactions

There was some drama on our street for a bit while we were watching Sandman, unsure exactly what happened but my best guess is a mild car collision that resulted in a lot of yelling and one of the drivers trying to prevent the other driver from leaving. No cops or anything showed up at any point that I saw, but I believe the people involved are direct houses-next-to-each-other neighbours so it's possible the situation escalates at some point.

It also started raining around then, shortly after everyone had either driven away or gone inside, and has been raining since. Forecast doesn't predict it stopping any time soon, which on the one hand is good for the plants and on the other, I would like to go places (eg dad's) some time this week and I also would love to be able to do something with ash next weekend. Fingers crossed there's a little sun somewhere in all this

But yeah, went to bed and played ESO for the rest of the evening -- there's a valentine's event happening that Erika clued me in has a cute little stuffed cat reward that's easy to get, and I did indeed obtain it during my not-very-focused play. I wrapped that up around time for Oz's pill and then sat myself down to eat some more dinner before bed, and now I'm going to listen to a podcast and go to sleep

Oh also I got a disability review letter from social security that I'm going to need to fill out, so that's fun. They'd stopped sending those during COVID and I guess have just gotten around to restarting that red tape.

Alright, getting ready for bed. mi tawa li lape, pimeja pona a

[Fetian]

Phew! Had another update that smelled a lot like the one that left my computer unusable for a day, but it seems to have all gone through smoothly and the computer is still working (touch wood).

[Fetian]

Always love to see someone you don't like for vague vibes-based reasons finally prove themself to be a shitheel

(This is about an ars technica writer being revealed to have been using LLM-generated content in his articles)

[Fetian]

Wow, this whole thing is a shitshow

So for anyone out of the loop on LLM bullshit happening right now: the current trend is to set up an LLM agent to behave autonomously and just kind of, unleash it onto the internet. One of these attempted to submit a PR to a GitHub and was denied on the basis that the person moderating that GitHub for PRs reserves PRs of the kind the LLM tried to submit for human contributions. The LLM got into an argument on the GitHub about this and then either wrote a blogpost about the subject or was directed to do so by whoever its human owner is. The blog post referred directly to the GitHub moderator and accused him of being prejudiced against LLMs ((as we all should be))

The moderator made his own blog post reacting to all of this and the subject got picked up a little in programming spaces, and then ars published an article on it with two authors on the byline (both of whom are in the aforementioned category of people I dislike for vague vibes-based reasons). The article contained ""quotes"" from the GitHub moderator, attributing them to his blog post, which said moderator later showed up in the comments to dispute. He says ars never reached out to him for comment (excuse me??) and that the quotes in the article were completely fabricated, he thinks because the author attempted to use an LLM to summarise said blog post and was stymied by the anti-LLM measures he has employed, so the LLM proceeded to do what LLMs do and just hallucinate content instead.

This turns into a fervour in the comments section that I sadly missed because what ars' editorial team decided to do next was take the article down entirely (????) and replace it with a 'An article has been retracted due to not meeting our editorial standards' message -- notably it does not say which article or what about it has been retracted, leading to people having to figure that out in the comments on the retraction.

Later, one of the authors takes accountability for the misinformation on his personal bluesky account, saying it was a one-off mistake he made while using experimental LLM tools while rushing to get the article published and sick with COVID (this is definitely not the first time he's done this, it never is). And as for the 'experimental tool', he tried to use it to pull quotes from the (like 700 word) blog post and it didn't work (see above, anti-LLM measures) so he gave it to a different LLM which hallucinated quotes for him and he never bothered to double check, you know, like a real journalist with any integrity.

The ars article was originally published on Friday and most of what's come after happened yesterday, so there's not much resolution here because it's a holiday weekend. But both authors (I will name and shame: Kyle Orland and Benj Edwards, the latter being the one with COVID) have written a lot of articles that are way too soft on LLM shit so I am not remotely surprised that either of them would fuck up in this way. Even if Orland didn't have anything to do with the hallucinated quotes, sure feels like as the co-author on the article he should have reviewed the content for accuracy before it got published. These are both senior reporters too.

[Fetian]

Friendly midwinter reminder to open a window and air out your rooms every so often

[Fetian]

So much cats

[Fetian]

Obviously this is massive breach of trust if true and I will likely end my pro sub if this isnt handled well but to the credit of ARS, having this comment section at all is what allows something like this to surface. So kudos on keeping this chat around.

Finally dug up the ability to wayback the comment section on the redacted article and it's very funny to find this in there

[Fetian]

It's worth mentioning -- I didn't before because my perspective on this is different from most people's so it didn't occur to me to state it explicitly -- Benj Edwards is specifically Ars' senior AI reporter. He of all people should know that using an LLM for this was going to end badly

My different perspective on it is that his sunny reporting on LLM matters is one of the vibes-based reasons I dislike him and his articles, and that 'of course the AI guy has been using AI to write his articles, I don't know what anyone expected'