this post was submitted on 10 Nov 2024
372 points (80.6% liked)
Technology
59415 readers
2914 users here now
This is a most excellent place for technology news and articles.
Our Rules
- Follow the lemmy.world rules.
- Only tech related content.
- Be excellent to each another!
- Mod approved content bots can post up to 10 articles per day.
- Threads asking for personal tech support may be deleted.
- Politics threads may be removed.
- No memes allowed as posts, OK to post as comments.
- Only approved bots from the list below, to ask if your bot can be added please contact us.
- Check for duplicates before posting, duplicates may be removed
Approved Bots
founded 1 year ago
MODERATORS
you are viewing a single comment's thread
view the rest of the comments
view the rest of the comments
Wonder where chatgpt will get its training data in the future, as it's known not to extrapolate well. Where will it learn new frameworks, languages, ... from?
Its going to starve itself.
Let it die
I doubt it ever scraped SO, otherwise all the answers would be smth along the lines: "I cannot answer this question due to low quality effort!" closes browser window
The documentation?
The docs. It's what it does now a lot of the time I've noticed.
Auto generated docs since devs don't document?
Chatgpt, look at this repo and write docs
Somebody already did that but it wasn't with chat GPT and honestly the docs were fine.
It didn't do that thing that a lot of humans do when writing documentation which is just declare that something is true without explaining why it is true. So you end up in random PHP like land, when things just work like that okay.
Honestly it's petty good about doing that. Already had similar tooling options but it does a generally good job of making docs for non devs assuming good naming are used in the methods
Yeah the smaller the project the less effective this is.
But even learning from the source code is pretty effective.
That works when the docs are good and clear. Otherwise, we'll have to revert to communicating with each other for brief periods while the chat-bots train themselves on the new data.
A lot of models are being trained on “synthetic” data now, right?
Even when a parrot learns to parrot a parrot, the first parrot still has to be taught.
Armies on paid personal generating content?
I see absolutely no problem with that.