What's in the News? 2.0

[Missy]

Companies required to accept unions or hold free and fair union votes, no tampering or else

[inkybus]

Companies trying to union bust? What is this, the 1890s?

[GloomCookie]

Most companies, even those with active unions, hate dealing with them because collective bargaining makes it harder to just abuse employees. When you have a single employee against management, management will win every single time. But if you have a union rep who is there to back you up and tell management the collective bargaining agreement protects the employee, then management has to be more careful.

My dad used to work as an HR manager, and he kept a good and open dialog between himself and the union he worked with. He would often tell me that one of his jobs was to keep management aware of what they legally could and could not do when it came to employees, while at the same time would deal with employees who thought because they were in the union that they were untouchable. Sometimes though even the union rep would just quietly pull the employee aside and explain in no uncertain terms how fucked they were.

[TheGlyphstone]

Companies trying to union bust? What is this, the 1890s?

Well WoTC did just send in the Pinkertons a few months ago...

[Al Terego]

Well WoTC did just send in the Pinkertons a few months ago...

This?  It wasn't union busting.

[TheGlyphstone]

This?  It wasn't union busting.

I'm making a joke about how inkybus said its the 1890s all over again.

[Al Terego]

Ah, sorry.

[Missy]

Most companies, even those with active unions, hate dealing with them because collective bargaining makes it harder to just abuse employees. When you have a single employee against management, management will win every single time. But if you have a union rep who is there to back you up and tell management the collective bargaining agreement protects the employee, then management has to be more careful.

My dad used to work as an HR manager, and he kept a good and open dialog between himself and the union he worked with. He would often tell me that one of his jobs was to keep management aware of what they legally could and could not do when it came to employees, while at the same time would deal with employees who thought because they were in the union that they were untouchable. Sometimes though even the union rep would just quietly pull the employee aside and explain in no uncertain terms how fucked they were.

that's pretty much how it works in unionized environments.

Sadly unions are at their weakest in recent times, even mine, which is stable, is subject to certain limitations.

[Sara Nilsson]

https://www.the-sun.com/entertainment/8938062/arleen-sorkin-dead-harley-quinn-days-of-our-lives/

Arleen Sorkin passed away, aka Harley Quinn for you youngesters.. or Calliope Jones in Days of our lives for the slightly older folks.

[TheGlyphstone]

Someone find Mark Hamill and seal him up in one of those protective bubbles...

[Beorning]

Crap. Is Pope Francis really that ignorant of Eastern European history?

https://www.reuters.com/world/europe/pope-francis-draws-criticism-extolling-russian-imperialist-tsars-2023-08-28/

The "enlightened empire" of Catherine the Great. Riiiiiiiiiiiiiiiiiiiiiiight.

[Chulanowa]

Crap. Is Pope Francis really that ignorant of Eastern European history?

https://www.reuters.com/world/europe/pope-francis-draws-criticism-extolling-russian-imperialist-tsars-2023-08-28/

The "enlightened empire" of Catherine the Great. Riiiiiiiiiiiiiiiiiiiiiiight.

I mean just the phrase "enlightened empire" is pretty funny.

[GloomCookie]

"Hmm these Tsars are pretty bad people. What about this Catherine lady? Oh she's Catherine the Great? She must have been ok then."

[Oniya]

Never mind actual history - I don't know a single college-age individual that isn't aware of the 'legends' about Catherine the Great, which I would think would make the head of the Catholic Church shy away from extolling her virtues.  (Although the historical verdict on that legend is 'neigh'.)

[Missy]

I almost forgot to share this.

You've been buster, buster (I just can't help that snarky line)

Seriously it's about gorram time we strengthened unions in this damned country.

[Oniya]

Er - no you didn't.

Companies required to accept unions or hold free and fair union votes, no tampering or else

[Missy]

gee, I've been so exhausted from work; the last two and a half weeks have been just brutal honestly. School just started and not only are we short staffed to begin with, the help we get is only so good on account they're kids who've never had a job before and you almost have to literally hold their hands every step of the way sometimes.

[Vekseid]

Doug Lenat has died. He was the founder of Cyc. As LLMs seem to be requiring ridiculous amounts of additional power for comparatively little additional gain, I think Lenat's work is going to get a lot more attention in the coming decade.

[GloomCookie]

I would like some context, please. I'm not familiar with LLMs.

[Al Terego]

Large Language Models

Also:
R.I.P Jimmy Buffett.

[Vekseid]

I would like some context, please. I'm not familiar with LLMs.

Large Language Models like ChatGPT, Bard, Claude, and Llama.

They are built on top of what is effectively a dictionary and a prediction engine (the Transformer - the 'T' in GPT). What reasoning these are capable of, beyond what the dictionary can immediately provide, needs to be beaten into the model with comparatively ludicrous amounts of training.

Meanwhile adjusting these models can easily lead to hard-to-predict results. ChatGPT has suffered some famous collapses in its mathematical performance, for example.

[Rinzler]

Meanwhile adjusting these models can easily lead to hard-to-predict results. ChatGPT has suffered some famous collapses in its mathematical performance, for example.

Most of us learnt the basics of arithmetic as kids by counting our fingers. I can certainly see why that would be a problem for AI. 

[Oniya]

Nobody fed it the Principia Mathematica?



https://en.wikipedia.org/wiki/Principia_Mathematica

[GloomCookie]

My experience with how AI learns is really, really dated given we only touched on it when I was in college, but based on my understanding, it's trying to force a square peg into a round hole.

Neural learning algorithms such as ChatGPT tend to map their responses based on brute force learning to try and predict based on weights what it is you want it to deliver. The weight of each word plays a part in how likely it is to appear, but there's more to it than just the number of times that word appears, or else ChatGPT would output over and over "THE THE THE THE THE THE BE BE BE BE TO TO TO" etc. and so it's using context clues and its predictive algorithm to take what you previously asked about and parsing together the words it thinks you want. There's no actual thought process, it's just using statistical analysis to link the words together into a coherent string.

Mathematics fucks with that because it's an entirely different operation entirely. The model built to use statistics to put together English words and phrases isn't geared for the comparatively simpler operation of following direct order of operations for mathematics. When you hand ChatGPT a simple equation like 2 + 2 = ?, it takes that and puts it together based on the statistical probability and says "Based on this, the answer is probably 4" because it's been taught that 4 is the likely answer you want. If you ask it to complete the phrase "It's raining ____ and ____" it will put together that, statistically, it should be "It's raining cats and dogs" but it doesn't have statically knowledge for the answer to "What's the square root of 2?"

So I can sorta see why ChatGPT doesn't handle math too well. Different tools for different tasks.

[Al Terego]

https://apnews.com/article/saudi-arabia-death-sentence-twitter-a2b5549806605d1d21f332ac4c36e43f