Comment by sixhobbits
1 day ago
(author here) I did feel kinda bad about it as I've always been a 'good' HNer until that point but honestly it didn't feel that spammy to me compared to some human generated slop I see posted here, and as expected it wasn't high quality enough to get any attention so 99% of people would never have seen it.
I think the processes etc that HN have in place to deal with human-generated slop are more than adequate to deal with an influx of AI generated slop, and if something gets through then maybe it means it was good enough and it doesn't matter?
That kind of attitude is exactly why we're all about to get overwhelmed by the worst slop any of us could ever have imagined.
The bar is not 'oh well, it's not as bad as some, and I think maybe it's fine.'
well, he was arguing that it's not worse than 99% of the human slop that gets posted, so where do you draw the line?
* well crafted, human only? * Well crafted, whether human or AI? * Poorly crafted, human * well crafted, AI only * Poorly crafted, AI only * Just junk?
etc.
I think people will intuitively get a feel for when content is only AI generated. If people spend time writing a prompt that doesn't make it so wordy, and has personality, and it OK, then fine.
Also, big opportunity going to be out there for AI detected content, whether in forums, coming in inmail inboxes, on your corp file share, etc...
Did you?
Spoiler: no he didn't.
But the article is interesting...
It really highlights to me the pickle we are in with AI: because we are at a historical maximum already of "worse is better" with Javascript, and the last two decades have put out a LOT of javascript, AI will work best with....
Javascript.
Now MAYBE better AI models will be able to equivalently translate Javascript to "better" languages, and MAYBE AI coding will migrate "good" libraries in obscure languages to other "better" languages...
But I don't think so. It's going to be soooo much Javascript slop for the next ten years.
I HOPE that large language models, being language models, will figure out language translation/equivalency and enable porting and movement of good concepts between programming models... but that is clearly not what is being invested in.
What's being invested in is slop generation, because the prototype sells the product.