textninja

The quality is noticeably worse than 4 but the tighter feedback loops and lower expectations make up for it most of the time. So I’m inclined to agree; speed matters.

Must be in the trough of disillusionment. The tech is truly useful though so we’ll reach the plateau soon enough.

It appears to summarize the subtitles alone. I asked it to summarize a video that didn’t have subtitles (silent film) and it refused.

Edit: I missed some of the context, comment above is re you-tldr, not kagi which I haven’t tried.

Chrome extension maybe?

I really like this idea. It’s basically applying similar principles as are used in image based nets - i.e. sliding window convolutional kernels - to text.

I’m not fond of the provocative title because prompting is easy and only getting easier; the advice seems to be predicated on the use of relatively deficient LLMs. I don’t doubt there will still be operator skill involved, but I anticipate the state of the art for LLMs ability to adapt to “bad” prompts will outpace our ability to learn to prompt them effectively.

Disclaimer: I watched the video but didn’t read the paper.

I call bullshit. There will be bigger and better models. The question is not whether big companies will invest in training them (they will), but whether they’ll be made available to the public.

> Soon enough no one will even know how to write, just read, because ChatGPT will write everything.

Nonsense, writing is easy! Just dictate some rough instructions to a GPT agent and copy/paste its response.

I’m being facetious, of course - writing is thinking, so I don’t think it’s necessarily going anywhere, though AIs can obviously augment or replace a lot of the busywork. Where ChatGPT is used to generate content absent thoughtful prompting, the stuff it spits out will largely be regarded as spam.

It took some coaching to get it to use the terse style I prefer, but here’s what 3.5 came up with in a conversation where I first asked about Rust’s support for regular expressions:

  use regex::Regex;
  
  fn strip_suffix_digits(s: &str) -> &str {
    let re = Regex::new(r"_\d+$").unwrap();
    re.replace(s, "").as_ref()
  }

As someone who does not program in Rust I find that quite readable, though I haven’t run it and can’t comment on how it compares outside of LOC. It does depend on the regex crate though.

> my instinct is to reach for regexp captures, the solution it came up with is probably a lot faster and easier to read and avoids "now you have two problems”.

I don’t write Rust but I think it’s best to trust your instinct here. “Now you have two problems” is a humorous quip and not practical coding advice. Using a regular expression to strip trailing digits from a string will surely result in code that is shorter and more readable than the alternatives, and it will probably be more correct too.

There’s a simple solution to that problem.

Programming isn't just coding.

> This article seems overly dramatic

The article: “I arrive in Taiwan brooding morbidly on the fate of democracy.”

> The design is not consistent (pages often don’t look the same).

I don’t think site-level consistency is as important or prevalent as you’d think.

I’m a programmer, not a philosopher, so I never actually heard of him, but from his Wikipedia entry I learned that “he rejected the notion of inalienable rights advanced by Locke and most contemporary capitalist-oriented libertarian academics,” and let’s just say it didn’t say anything about him getting fired from his job.

That’s “hacking” in the same way as changing the page number parameter in a URL is hacking ;)

Writing is thinking

I’ll admit to using the words framework and library interchangeably when referring to React in my day to day, but what distinguishes React from frameworks to me is the fact it doesn’t tie itself to a particular rendering context, i.e. it is flexible and unopininated and relies on wrappers and rendering libraries to actually function.

Marked. And I wouldn’t bet against you, either.

> And this is the company leading the way on AI.

Well it is called artificial intelligence. But in all seriousness, they no doubt have brilliant people working there. I don’t however consider them to be a tech company, at least in the traditional sense. Scientific? Absolutely, but their nonprofit roots are really showing through.

I’m aghast that using a general purpose AI for general purposes is treated as a malicious act and censured as such. “No! Not those tokens!!”

Sorry to be obtuse, but it’s not damning at all because it’s a library, not a framework.

That’s “hacking” in the same way as pressing a button on the elevator is trespassing.

Not to mention the LLM equivalent of Trojan horses

Most likely it’s just a goofy dev plugin that autoinjects the same DAN prompt everyone uses.

> You say you're 'arguing between the lines.' Do you code between them, as well?

Wouldn’t it be impressive if I did?

> aka the brainwashed AI that reinforces their political and cultural biases

I love seeing IronyAI called out like this.

> Which of course implies that I successfully inferred your accelerationist libertarianism from context. :)

I mean, I did use 3 out of 8 points to rag on ethicists.

> Yes, this is basic Logic 101 stuff. Not even higher-order logic.

Best practice is to forget everything you ever learned in a 101 class. And don’t bring logic to a word fight.

The hilarious thing is that we do say a car has power equivalent to that of, say, 887 horses, but when it’s about humans it suddenly becomes nonsensical to make a comparison.