istewart

joined 9 months ago
[–] istewart@awful.systems 6 points 3 weeks ago (1 children)

I feel like I've seen chud weirdos ranting about seed oils suppressing testosterone levels, but I could be hallucinating

[–] istewart@awful.systems 5 points 4 weeks ago

Hey, at least the professional Elon haters are actually doing their jobs, unlike the man himself

[–] istewart@awful.systems 12 points 1 month ago

I am certain that many of those ignorant of the history (or even were there for it, like DHH) would still argue that Google deserves credit because of the V8 JavaScript engine. But I continue to doubt that further promulgating JavaScript was a net positive for the world.

[–] istewart@awful.systems 12 points 1 month ago

Wow, coiners tried to parasitize YIMBY while I wasn't paying attention?

[–] istewart@awful.systems 8 points 1 month ago

I picked up a modern Fortran book from Manning out of curiosity, and hoo boy are they even worse in terms of trend-riding. Not only can you find all the AI content you can handle, there's a nice fat back catalog full of blockchain integration, smart-contract coding... I guess they can afford that if they expect the majority of their sales to be ebooks.

[–] istewart@awful.systems 4 points 1 month ago

I'm mildly surprised that music of any kind is what's getting uploaded to something called Deezer

[–] istewart@awful.systems 9 points 1 month ago

As I noted on the YouTube video, this is doubly heinous as a lot of CA community college instructors are "freeway flyers" - working at multiple campuses, sometimes almost 100 miles apart, just to cobble together a full-time work schedule for themselves. Online, self-paced, forum-based class formats were already becoming popular even before the pandemic, and I've been in such classes where the professor indicated that I was one of maybe 3 or 4 students who bothered to show up to in-person office hours. I have to wonder if that will end up being a hard requirement at some point. The bottom rung on the higher-education ladder is already the most vulnerable, and this just makes it worse.

[–] istewart@awful.systems 6 points 1 month ago (2 children)

I have to agree. There are already at least two notable and high-profile failure stories with consequences that are going to stick around for years.

  1. The Israeli military's use of "AI" targeting systems as an accountability sink in service of a predetermined policy of ethnic cleansing.
  2. The DOGE creeps wanting to rewrite bedrock federal payment systems with AI assistance.

And sadly more to come. The first story is likely to continue to get a hands-off treatment in most US media for a few more years yet, but the second one is almost certainly going to generate Tacoma Narrows Bridge-level legends of failure and necessary restructuring once professionals are back in command. The kind of thing that is put into college engineering textbooks as a dire warning of what not to do.

Of course, it's up to us to keep these failures in the public spotlight and framed appropriately. The appropriate question is not, "how did the AI fail?" The appropriate question is, "how did someone abusively misapply stochastic algorithms?"

[–] istewart@awful.systems 11 points 1 month ago

Would you invest in commercial real estate, knowing there was a non-zero chance your tenants might come in one day to discover a thoroughly intoxicated JD Vance in a compromising position with the break-room furniture?

[–] istewart@awful.systems 9 points 1 month ago (1 children)

Mesa-optimization... that must be when you rail some crushed-up Adderall XRs, boof some modafinil for good measure, and spend the night making sure your kitchen table surface is perfectly flat with no defects abrasions deviations contusions...

[–] istewart@awful.systems 15 points 1 month ago* (last edited 1 month ago)

couldn't help myself, there are seldom more perfect opportunities to use this one

[–] istewart@awful.systems 11 points 1 month ago (1 children)

Another thread worth pulling is that biotechnology and synthetic biology have turned out to be substantially harder to master than anticipated, and it didn't seem like it was ever the primary area of expertise for a lot of these people anyway. I don't have a copy of any of Kurzweil's books at hand to look at his predicted timelines for that stuff, but they're surely way off.

Faulty assumptions about the biological equivalence of digital neural network algorithms have done a lot of unexamined heavy lifting in driving the current AI bubble, and keeping the harder stuff on the fringes of the conversation. That said, I don't doubt that a few refugees from the bubble-burst will attempt to inflate the next bubble on the back of speculative biotech, and I've seen a couple of signs of that already.

view more: ‹ prev next ›