> the best alternative to rigorous specification is human review. human review doesn't scale well to the volumes of output produced by language models. to make matters worse

when the business model is selling more tokens you get such per serve ice times that lead to “more” thinking, engagement baiting, fluffy narratives, and straight up dark patterns

but most of what humans create isn’t novel, esp in saturated areas like art. so even LLMs can “just do that” they’re very capable at outputting worthwhile materisl

"most of" being the key words.

straight up lying and doomer boomer mentality, sad and gross.

very consistent correlations between migrants being net positive economic providers, more law abiding then native citizens, and just straight up regurgitating racist myths

mass deportations won’t solve anything economic when it’s the same admin hamstringing prices and rates with tariffs, war, and defacto support of the oligopoly

i really wish basement dwellers like you though of something original now and then

its afraid

i’m not trying to dismiss this article entirely as i did get value around from discussion around verifying

but this coming from a VC firm invested in token producing companies and quoting the “genius” behind openclaw makes me think this is just an psyop of over engineering to get ppl to blindly spend tokens on the LLM slot machine

Not really; looping is a requirement when there's an imperfect generator: LLMs. I read the article some days ago and included the citation in my own.

For a longer explanation why, here's mine:

https://memseek.ai/blog/the-loop-has-always-been-there/

ps: not affiliated with a16 or similar

Put another token in the slop machine!

i have unlimited tokens being a large corp so i’m a bit detached from billing and even general best practices for promoting

but the incentives of these companies to become profitable at any cost slipping into entire new types of dark patterns around token based billing seems gross

- charging for injected prompts and cot tokens

- changing default thinking effort to be higher

- training models to give longer winded answers that don’t say anything more of substance

- refusing to fulfill a request and still charging you

i wonder if you could ever just charge based of each user message and it so how breaks even across short and long replies

i’m sure everyone replying to this thread has a level headed and unbiased assessment of themselves

That's what makes us so smart!

substack has devolved into the same thing. opening the app first thing you see if a infinite scroll of twitter like takes

and i honestly would rather have a reddit lite site like hacker news where there appears to be at least a diversity of opinion than the pseudo-intellectualization cjing occurring among the exact same sort of AI cockgobblers on LW

> and i honestly would rather have a reddit lite site like hacker news where there appears to be at least a diversity of opinion

Why? Isn't that available on literally every single other tech comment section on the Internet? Like what distinguishes these angry comments from the ones on Techcrunch or Ars or Reddit subs that tolerate AI? If I wanted mindless diversity I'd just have the LLM generate random perspectives.

and how does the public view AI tech workers?

if you’re visiting major cities and need basic clothing you’re probably shopping at the some of the same name brand outlets or lines

But you will have a much better idea what to buy, when you have first-hand experience with the weather and are surrounded by locals who are adequately prepared.

interesting how much overlap onebagging has with ultralight backpacking