15 lines
818 B
JSON
15 lines
818 B
JSON
{
|
||
"HubID": "5578",
|
||
"Date": "10/14/2025",
|
||
"HubTags": [
|
||
"External Platform Posts",
|
||
"Future Map"
|
||
],
|
||
"Contacts": "",
|
||
"Companies": "",
|
||
"File": "",
|
||
"Image": "",
|
||
"Summary": "<p>A recent study found that being rude to LLMs actually improved their accuracy. That may not be surprising, considering these models are trained on the Internet—where algorithms often reward impolite, emotionally charged content because it drives engagement. As the web becomes increasingly flooded with bots, AI-generated slop, clickbait, and ragebait, we’re training our models—and ourselves—to reflect the worst parts of online behavior. People are starting to recognize this, and efforts are emerging across AI and content fields to help correct it.</p><p>https://arxiv.org/abs/2510.04950</p>",
|
||
"Notes": ""
|
||
}
|