robjwells

joined 3 days ago
[–] robjwells@awful.systems 3 points 5 hours ago

Even 14-year-old games aren’t safe from the slop machine, as Torchlight II may get new updates using AI:

Given the age of the game, restructuring its complex underlying code presents many challenges. However, new AI-assisted coding tools have opened up more possibilities for reviewing the code and addressing longstanding issues.

[–] robjwells@awful.systems 4 points 16 hours ago

Native English speakers might recognise it as Storm of Steel; for the unfamiliar it’s Jünger’s memoir of WWI and arguably a celebration of war.

Jünger’s an interesting character. He had different politics to the Nazis but it would be fair to call him a fascist in the period from 1918-45. After the war he revised some of his works to step away from some of his previous views, and had a big falling out with his private secretary Armin Mohler (an unrepentant fascist) over it. By the 1980s he was publicly denouncing the “ideology of war” contained in Storm of Steel.

If anyone is interested, I highly recommend the series on Mohler’s book The Conservative Revolution in Germany by the anti-fascist podcast The Empire Never Ended. First ep on YouTube. Jünger is discussed in the episode on the National Revolutionaries.

[–] robjwells@awful.systems 11 points 1 day ago (3 children)

Nvidia claims deployment could shrink from six months to as little as 14 days at a prepared site. That excludes the not insignificant matter of getting power, connectivity, and other site infrastructure ready first.

Pairs well with that report about Amazon having billions of dollars worth of GPUs in storage because they don’t have DCs to put them in. At least these come pre-packaged in a container you can just park somewhere!

"At 2 gigawatts, DSX can save 54 million litres of water a day," Huang claimed, perhaps thinking of those environmental protesters.

That’s the equivalent daily water use of 375,000 people (UK figures). Obvious next question is “yes but how much water are you still using after that?”

[–] robjwells@awful.systems 10 points 2 days ago* (last edited 2 days ago)

I want to point out that SMW Central doesn't ban people for being pro-GenAi, nor are they barred from anything just for being pro-GenAi. They simply can't upload content to the site that was created with GenAi. Despite this, GenAi users believe that, because SMW Central has rules against GenAi content, SMW Central must feel the need to "control or dictate" what they do.

This context transforms the statement substantially. When this group of individuals says "nobody owns ROM hacking," what they mean is that they should be allowed to publish their work wherever they want without consequence.

Kind of reminds me of the people who, around the time the (UK) smoking ban was introduced, made "It's my right to smoke anywhere I want" their whole personality.

20
submitted 2 days ago* (last edited 2 days ago) by robjwells@awful.systems to c/techtakes@awful.systems
 

78% "unhappy" or "complacent" at work. 59% avoiding AI use.

Note though that far fewer people answer the survey than they once did: 30,000 this year vs a peak of 100,000 in 2018. And the questions above were answered by less than 2/3 of the total respondents.

Edit: Just realised this was published over a week ago, sorry, it was the first time I saw the banner on SO.

[–] robjwells@awful.systems 3 points 3 days ago

Good luck with your exams.

[–] robjwells@awful.systems 4 points 3 days ago* (last edited 3 days ago)

Is this just a slightly altered repeat of when LLM companies (or was it some boosters?) argued they were so dangerous and so potentially profitable the best thing would be for the government to buy 50% - 1 of their shares?

an additional 2% added at the board’s discretion for every 50% increase in the company’s value above a set baseline

Exceedingly optimistic to suggest that they haven't already peaked in private valuation (perhaps apart from any IPO pop).

[–] robjwells@awful.systems 8 points 3 days ago* (last edited 3 days ago)

They got one of my favourites, Medhi Sadaghdar (ElectroBoom), and yes in a video released on the 10th.

CAIS seem like an "AI safety" laundromat, in the sense they try to put a reasonable face on the paperclip maximiser stuff. They were the ones who organised the "AI safety statement" in 2023, which features lots of signatories who I'm sure just thought "oh yes I don't want anything bad to happen because of AI", not "If we don't give Yudkowsky the reins we're doomed to become office supplies".

[–] robjwells@awful.systems 2 points 3 days ago

Very disappointing, even if I was only ever a non-registered lurker. Gave in and DNS-blocked it and HN this morning (I had a HN-filtering userscript which in hindsight was just cope).