FredFig @ FredFig @awful.systems

Posts

0

Comments

94

Joined

1 yr. ago

I think it's a piece in the long line of "AI means A and B, and A is bad and B can be good, so not all AI is bad", which isn't untrue in the general sense, but serves the interest of AIguys who aren't interested in using B, they're interested in promoting AI wholesale.

We're not in a world where we should be offering AI people any carveout; as you mention in the second half, they aren't interested in being good actors, they just want a world where AI is societally acceptable and they can become the Borg.

More directly addressing your piece, I don't think the specific examples you bring up are all that compelling. Or at least, not compared to the cost of building an AI model, especially when you bring up how it'll be cheaper than traditional alternatives.

At the risk of doing some "founder mode" idiot's homework for them, impermanence is inherent to a lot of artforms, and I can see some insane and vague pitches to use quantum to "capture the magic moment". Or maybe they tie it back into NFTs with quantum technology that comes up with every variant of the bored chimpanzee at once.

1. The inability to objectively measure model usability outside of meme benchmarks that made it so easy to hype up models have come back to bite them now that they actually need to prove GPT-5 has the sauce.
2. Sam got bullied by reddit into leaving up the old model for a while longer, so its not like its a big lift for them to keep them up. I guess part of it was to prove to investors that they have a sufficiently captive audience that they can push through a massive change like this, but if it gets immediately walked back like this, then I really don't know what the plan is.
3. https://progress.openai.com/?prompt=5 Their marketing team made this comparing models responding to various prompts, afaict GPT-5 more frequently does markdown text formatting, and consumes noticeably more output tokens. Assuming these are desirable traits, this would point at how they want users to pay more. Aside: The page just proves to me that GPT was funniest in 2021 and its been worse ever since.

We're at the point of 100xers giving themselves broken sleep schedules so they can spend tokens optimally.

Inevitably, Anthropic will increase their subscription costs or further restrict usage limits. It feels like they're giving compute away for free at this point. So when the investor bux start to run dry, I will be ready.

This has to be satire, but oh my god.

Half of these are people using GPT to write a rant about GPT and the other half are saying "skill issue", it's an entirely different world.

affecting less than 5% of users based on current usage patterns.

This seems crazy high??? I don't use LLMs, but whenever SaaS usage is brought up, there's usually a giant long tail of casual users, if its a 5% thing then either Copilot has way more power users than I expect, or way less users total than I expect.

Looks like it's been downranked into hell for being too mean to the AI guys, which is weird when its literally an AI guy promoting his AI generated trash.

This caused me to reconsider something. I had kinda assumed that everything sucks because the bar of quality for software is so low, and that's pulling it down for every other field now that software proliferated into eating the world.

But I didn't examine that the relationship could work in both directions. Software sucks some of the time, but it doesn't excuse shit like how Crowdstrike can still be in business, and we should probably look into what's caused us to develop the attitude about not caring that shit is shit, just because the shit salesmen told us it'll be less shit in the future.

I don't like to speculate, but it seems tptacek is not happy about it for some reason: https://news.ycombinator.com/item?id=44291623

We already have AI summarized Dune, it's called WH40k.

https://www.gauntletai.com/

10 weeks of 100h work weeks so you can have a 98% (publically disclosed) chance of being denied a Golden Ticket to the AI factory.

This is very weird but not particularly notable, other than that these guys have apparently been YC funded in 2017, and I can't find anything about the company in the directory: https://www.ycombinator.com/companies?batch=Summer+2017... until I looked at the CEO's name. Lambda School Bloom Institute GauntletAI's latest pivot is asking for 1000 hours of voluntary unpaid labour.

e: apologies for the vaguepost, I could've sworn I put down the name, but in any case, it's the return of PG's favourite guy not named Sam Altman, Austen Allred

Being paid to reddit has to be the most pathetic thing you can do, and I say that as someone who once reddited unpaid.

This almost reads like tptacek doesn't understand why lucidity's piece a year ago was so effective and tried to write it from the opposite angle by punching down instead of punching up.

I'd have thought that a guy who writes on the internet like it's a competition sport could recognize the obvious problems of this, but maybe I'm just a vibe coding Youtuber.

Being the worst and the biggest studios aren't unrelated concepts, they're the biggest because they're the worst.

I don't get it, how is every one of the most touted people in the AI space among the least credible people in the industry.

Like literally every time its a person whose name I recognize from something else they've done, that something else is something I hate.

Reminds me something F.D. Signifier said on a music podcast.

Progressives are losing the cultural war in a lot of ways, but they'll always need us because we're the ones pushing the boundaries on art, and it turns out, no matter how ghoulish people want to act, everyone has genuine love of fucking awesome art. The true loss condition is being captured by the tools of the master.

Listen, I'm just saying that this never would've happened with hg.

If you're referring to genetic algorithms, those work by giving the computer some type of target to gun for that's easy to measure and then letting the computer go loose with randomly changing bits from the original object. I guess in your mind, that'd be randomly evolving the codebase and then submitting the best version.

There's a lot of problems with the idea of genetic codebase that I'm sure you can figure out on your own, but I'll give you one for free: "better code" is a very hard thing for computers to measure.

This was a year before the stuff with the Zizians happened too.

Hey that's unfair, us North Americans are as normal about bikes as we're normal about cars.