Hacker Newsnew | past | comments | ask | show | jobs | submit | SquareWheel's commentslogin

> Did vibe coding the business page inject it into ChatGPT's training data?

No, at least not directly. Inference does not train models. It is possible that OpenAI may separately collect the chat data, clean it, and feed it back into the model for future iterations. Or they could have extracted URLs for future indexing.

More likely though, I suspect, is your site just managed to be indexed naturally, and LLMs are very efficient at matching obscure data to relevant queries.


Interesting. Maybe someone could run bot farms that ask variants of the same question and subtly nudge the model by replying reasons why the model's recommended service A is inferior to service B. Or other forms of adversarial question answers sessions.

This is a pretty banal comment at this point. Open source is the term used in the LLM community. It's common and understood. Nobody is going to release petabytes of copyrighted training data, so the distinction between open source vs weights is a rather pointless one.


First you steal all the code, then you want to redefine the term? Is it never enough with you AI guys? Where's the humility, where's the good?


Sorry, too busy "stealing code" to answer right now.


lol


its still a pointed one.

"open source" keeps being redefined by people with wealth and power to restrict our computing rights.

eventually its just gonna be "proprietary microsoft code that runs on microsoft servers, but you can see a portion of the results"


"Open source" as a term has evolved due to its success. It wasn't some malicious attempt at redefining things from the technical elite. It was a natural shifting of language, as happens with all words, as it entered more common usage.

It's entirely reasonable that this colloquial understanding would be applied to new categories such as AI models. I'm sure it'll be applied to many other things that don't fit the OSD either. That's just language for you.


Tell this to the Allen project, Apertus Project, SmoLLM, etc, etc, etc


Single Page Applications use the History API to create a working back/forward history within the SPA. This will cause you to navigate away on use, and potentially lose data.


Why don't they just use # hash urls for their in-SPA page urls instead? That's what Elm and Elixir LiveView both do.


That sounds like a design failure.


Well, yes, but that's why it's behind an about:config flag and users should not enable it without understanding its effects.


It's not, I don't know why you'd think that.


Chromium does include an ad blocker for "intrusive ads". ie. Those using many resources, flashing, auto-playing sounds, or otherwise behaving badly.

https://support.google.com/chrome/answer/7632919


Manifest 3 explicitly enables ad blocking through the declarativeNetRequest API. It's trivial to do so, and many blockers exist in the Chrome Web Store.


Removing film crew, boom mics, and missed props from a scene would surely be useful to studios. It may even enable some shots that previously would have been impossible due to the positioning of cameras, etc.


Windows has a native (cloud-based) dictation software built-in[1], so there's likely less demand for it. Nonetheless, there are still a handful of community options available to choose from.

[1] https://support.microsoft.com/en-us/windows/use-voice-typing...


Bazzite KDE picked up my 8BitDo controller immediately, with no prior configuration. I didn't even have to manually pair the Bluetooth. I was very impressed.


Many people will likely be watching this show for the first time based on this thread. Please don't spoil major plot events for them.


This thread is already filled with spoilers, so it feels odd that you're singling mine out and especially given how long the show's been out.


Just in case you missed it, your quote was referring to encoding requirements. Decoding (eg. Netflix users) will have a different set of requirements. The situation will also improve over time as dedicated hardware encoders and decoders become available.

For the moment, I don't really mind if it requires more GPU power to encode media, since it only needs to happen once. I expect it will still be possible on a weaker card, but it would just take longer.


Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: